From 83c8a763a1bfdc733a2a531109486e6be03da5ef Mon Sep 17 00:00:00 2001 From: tegwick Date: Mon, 14 Sep 2026 22:04:44 +0200 Subject: [PATCH] Research deliverbles standard, concept on earned equity, data collection and digests added --- C3ResearchDeliverablesStandard.md | 846 ++++++++ concepts/CustomerEarnedEquityDesign.md | 1716 +++++++++++++++++ ...2f3d46cb1faeef612cc6d5ae1db68c36ec9145b9f9 | 1 + ...0fbcac5b7f9a18f45ae7e4fa10f799e7eda7d625a5 | 2 + ...fc61876397d1436f711c01266647d1fe7e818.json | 1 + ...ff4e00540c60a06b2a0fe633320119530f4af.json | 1 + ...dd67ca391c6b7c6c2cf0f58fac373581046c0.json | 1 + ...3b419c7f1c285eb7c68720c23e7879ba4fabd.json | 1 + ...74e60124bd8e46d82b68781048b7bbe1c10fa.json | 1 + ...9f828b0baee4592f1f80c28317a93b53472db.json | 1 + ...a56369dd71876699f1faeb65ea93b2932f9f8.json | 16 + ...ef525b7fd96bcce9b6f810ecd62a1af78e64f.json | 16 + ...2f9333cac55ef76c27952caaf1f0f7c4eedb0.json | 1 + digests/daily/2026-09-14.md | 120 ++ ...ative_AI_Strategy_for_Europe_Assessment.md | 810 ++++++++ ...ncyParticipationCapital_ResearchProgram.md | 931 +++++++++ 16 files changed, 4465 insertions(+) create mode 100644 C3ResearchDeliverablesStandard.md create mode 100755 concepts/CustomerEarnedEquityDesign.md create mode 100644 data/artifacts/3e331ed3352d4132bef85b2f3d46cb1faeef612cc6d5ae1db68c36ec9145b9f9 create mode 100644 data/artifacts/471e5e2ccda3459a0170200fbcac5b7f9a18f45ae7e4fa10f799e7eda7d625a5 create mode 100644 data/captures/34e27d084a98f664d939c0c58d5fc61876397d1436f711c01266647d1fe7e818.json create mode 100644 data/captures/5a9e9cf3db3bf6cd0e69e12252bff4e00540c60a06b2a0fe633320119530f4af.json create mode 100644 data/captures/6d006c52a910ae340af4d331637dd67ca391c6b7c6c2cf0f58fac373581046c0.json create mode 100644 data/captures/c5220d4543013a816b09b3696293b419c7f1c285eb7c68720c23e7879ba4fabd.json create mode 100644 data/captures/ccf19268ed24a3c37a6811bc83d74e60124bd8e46d82b68781048b7bbe1c10fa.json create mode 100644 data/ingestion-runs/512d6bd5fabd7c6dc107071fd2c9f828b0baee4592f1f80c28317a93b53472db.json create mode 100644 data/observations/d74dfd54aaae99d7f6ab4d02441a56369dd71876699f1faeb65ea93b2932f9f8.json create mode 100644 data/observations/da6fe5fde5eaa0b382c4f15ede0ef525b7fd96bcce9b6f810ecd62a1af78e64f.json create mode 100644 data/scheduler-runs/69f85cc3d7184dccaefbfebe4b52f9333cac55ef76c27952caaf1f0f7c4eedb0.json create mode 100755 digests/daily/2026-09-14.md create mode 100755 digests/topics/2026-09-14-1246_Transformative_AI_Strategy_for_Europe_Assessment.md create mode 100755 digests/topics/2026-09-14-1414_IntelligentEconomics_AgencyParticipationCapital_ResearchProgram.md diff --git a/C3ResearchDeliverablesStandard.md b/C3ResearchDeliverablesStandard.md new file mode 100644 index 0000000..5bb6288 --- /dev/null +++ b/C3ResearchDeliverablesStandard.md @@ -0,0 +1,846 @@ +--- +title: "C3 Research Deliverables Standard" +short_title: "C3ResearchDeliverablesStandard" +version: "0.1" +status: "initial" +artifact_type: "methodology" +scope: "research repositories" +--- + +# C3 Research Deliverables Standard + +**C3ResearchDeliverablesStandard** is a lightweight methodology for organizing research results into three distinct classes of durable deliverables: + +1. **Compendium** — what we know. +2. **Catalog** — what we have identified. +3. **Concepts** — what we propose or explore. + +The standard is intended for research repositories that continuously collect evidence, synthesize knowledge, design new ideas, and revise their understanding over time. + +Its purpose is to keep three fundamentally different epistemic activities separate: + +> **describing the world, inventorying the world, and designing possible changes to the world.** + +The three classes deliberately share the letter **C**, giving the method its name: + +> **C3 = Compendium + Catalog + Concepts** + +--- + +# 1. Why C3 exists + +Research repositories tend to accumulate heterogeneous Markdown files: + +- literature reviews; +- definitions; +- taxonomies; +- benchmark lists; +- hypotheses; +- design ideas; +- source summaries; +- proposed mechanisms; +- research notes; +- experiments; +- datasets; +- conclusions. + +Without a clear information architecture, these gradually become difficult to interpret. + +A polished speculative design may start to look like established knowledge. A catalog entry may acquire narrative interpretation and become difficult to process systematically. A compendium may become a dumping ground for every interesting fact. + +C3 separates durable research output by **epistemic purpose** rather than by topic. + +The standard asks three questions: + +> **What do we know?** + +> **What things have we identified?** + +> **What could we create or change?** + +Each question has one canonical home. + +--- + +# 2. The three deliverable classes + +## 2.1 Compendium — What do we know? + +The **Compendium** contains synthesized research knowledge. + +A compendium document explains the best current understanding of a subject based on available evidence. It may integrate theory, definitions, literature, empirical findings, competing explanations, historical development, uncertainty, unresolved questions and conclusions. + +A compendium is not merely a collection of sources. Its purpose is to transform source material into an increasingly coherent model of a subject. + +### Compendium question + +> **What is currently justified to believe about this subject?** + +### Typical artifacts + +- `IntelligenceCompendium.md` +- `EconomicsOfMachineIntelligence.md` +- `TransformativeAICompendium.md` +- `AgencyAndMotivation.md` +- `HistoryOfConsumerOwnership.md` + +### Compendium quality criterion + +A good compendium should make clear: + +- what is known; +- how strongly it is known; +- what is disputed; +- what evidence supports the current view; +- what remains unresolved. + +### Directory + +```text +compendium/ +``` + +--- + +## 2.2 Catalog — What have we identified? + +The **Catalog** contains structured inventories of identifiable entities, mechanisms, observations, patterns, sources or categories. + +A catalog is designed primarily for enumeration, comparison, retrieval, classification, coverage analysis and machine processing where useful. + +It should favor explicit fields and repeatable structure over long narrative prose. + +### Catalog question + +> **What relevant things exist, and how can they be systematically described and compared?** + +### Typical artifacts + +- benchmark catalog; +- model catalog; +- thinker catalog; +- mechanism catalog; +- policy catalog; +- indicator catalog; +- failure-mode catalog; +- experiment catalog; +- source catalog; +- use-case catalog. + +Examples: + +```text +catalog/ +├── AIBenchmarkCatalog.md +├── EconomicIndicatorCatalog.md +├── ConsumerCapitalMechanismCatalog.md +├── ParticipationCapitalFailureModeCatalog.md +└── ThinkerCatalog.md +``` + +### Catalog quality criterion + +A good catalog should be: + +- systematically structured; +- extensible; +- internally consistent; +- explicit about inclusion criteria; +- suitable for comparison; +- easy to update without rewriting the whole artifact. + +--- + +## 2.3 Concepts — What could we create or change? + +The **Concepts** area contains proposed mechanisms, candidate architectures, designed interventions, hypotheses developed into concrete models, and possible solutions. + +Concept documents are intentionally allowed to be speculative. + +Their purpose is not to claim that something already works. Their purpose is to make an idea sufficiently explicit that it can be criticized, modeled, compared, simulated, prototyped or tested. + +### Concept question + +> **What might work, and how exactly would it work if we built or adopted it?** + +### Typical artifacts + +- `CustomerEarnedEquityDesign.md` +- `IndividualParticipationCapitalAccount.md` +- `AdaptiveTransitionBuffer.md` +- `SovereignAIAssuranceNetwork.md` +- `ConsumerParticipationDividend.md` + +### Concepts quality criterion + +A good concept should make clear: + +- the problem it addresses; +- its mechanism; +- actors and incentives; +- expected benefits; +- assumptions; +- design parameters; +- trade-offs; +- failure modes; +- evidence level; +- how it could be tested. + +### Directory + +```text +concepts/ +``` + +--- + +# 3. The core epistemic distinction + +The three C classes represent different relationships to reality. + +## Compendium + +> **Models of the world** + +The compendium tries to explain reality. + +## Catalog + +> **Inventory of the world** + +The catalog identifies and structures relevant things in reality. + +## Concepts + +> **Possible changes to the world** + +Concepts propose interventions, mechanisms and designed alternatives. + +This distinction is the heart of C3. + +The same topic may legitimately appear in all three forms: + +```text +compendium/ +└── ConsumerOwnership.md + +catalog/ +└── ConsumerOwnershipMechanismCatalog.md + +concepts/ +└── CustomerEarnedEquityDesign.md +``` + +These files are not duplicates. They answer different questions. + +--- + +# 4. Directory standard + +A C3-compliant research repository SHOULD contain the following root-level directories: + +```text +repository/ +├── catalog/ +├── compendium/ +└── concepts/ +``` + +These are the **normative C3 deliverable directories**. + +A larger research repository MAY additionally contain supporting directories such as: + +```text +repository/ +├── catalog/ +├── compendium/ +├── concepts/ +├── data/ +├── digests/ +├── research/ +└── README.md +``` + +These supporting directories are not part of the C3 deliverable taxonomy itself. + +--- + +# 5. Supporting directories + +## 5.1 `research/` + +Contains working research artifacts such as dated investigations, exploratory notes, source reviews, literature scans, hypothesis exploration and preliminary analyses. + +Research files may be temporary or durable, but they have not yet necessarily been normalized into C3 outputs. + +Typical form: + +```text +research/ +└── 2026-09-14-transformative-ai-strategy-assessment.md +``` + +A research file may later update one or more C3 deliverables. + +## 5.2 `digests/` + +Contains time-bounded summaries of change such as daily intelligence digests, weekly research digests, monthly policy updates and benchmark change summaries. + +Digests answer: + +> **What changed during this period?** + +They are chronological rather than epistemically canonical. + +## 5.3 `data/` + +Contains raw or normalized research material intended for analysis, including CSV, JSON, YAML, benchmark observations, measurement exports and structured source metadata. + +Data supports C3 artifacts but is not itself a C3 deliverable. + +--- + +# 6. Placement decision rule + +When creating a durable research artifact, ask the following questions in order. + +## Question 1 + +**Is the main purpose to explain the current state of knowledge?** + +If yes: + +```text +compendium/ +``` + +## Question 2 + +**Is the main purpose to enumerate and compare identified things?** + +If yes: + +```text +catalog/ +``` + +## Question 3 + +**Is the main purpose to design, propose or explore something that does not yet exist in established form?** + +If yes: + +```text +concepts/ +``` + +## Question 4 + +**Is the artifact primarily an investigation, dated analysis, source review or working research result?** + +If yes: + +```text +research/ +``` + +until its results are incorporated into one or more C3 deliverables. + +## Question 5 + +**Is the artifact primarily a time-bounded summary of recent change?** + +If yes: + +```text +digests/ +``` + +--- + +# 7. The C3 research flow + +C3 is not intended as a rigid pipeline. Research can move between all three classes. + +A common pattern is: + +```text +research + ↓ +compendium + ↓ +concept + ↓ +catalog / experiment + ↓ +evidence + ↓ +research + ↓ +updated compendium +``` + +A more abstract formulation is: + +> **Understand → Invent → Enumerate → Test → Learn** + +The important principle is that **concepts do not silently become facts**. + +Evidence generated by testing a concept must return through research and eventually update the compendium. + +--- + +# 8. Concept maturity + +Concept artifacts SHOULD declare their maturity explicitly. + +Recommended maturity values: + +```yaml +maturity: + - hypothesis + - exploratory + - modeled + - simulated + - prototyped + - tested + - validated + - operational +``` + +Meaning: + +- `hypothesis` — the core causal idea has been stated. +- `exploratory` — the mechanism has been elaborated and alternatives considered. +- `modeled` — actors, incentives and expected dynamics have been formalized. +- `simulated` — the concept has been exercised in a simulation or analytical model. +- `prototyped` — a working implementation exists. +- `tested` — the concept has been tested against observations or experiments. +- `validated` — multiple lines of evidence support the concept under specified conditions. +- `operational` — the concept is used in a real system. + +A concept reaching `validated` or `operational` maturity may generate new compendium knowledge, but the concept document itself remains a concept artifact. + +--- + +# 9. Evidence maturity + +C3 artifacts SHOULD distinguish maturity of an idea from strength of evidence. + +A recommended evidence scale is: + +```text +E0 — intuition +E1 — historical analogue +E2 — observational evidence +E3 — quasi-experimental evidence +E4 — randomized evidence +E5 — replicated cross-context evidence +E6 — operational proof at meaningful scale +``` + +Example: + +```yaml +artifact_type: concept +maturity: modeled +evidence_level: E2 +``` + +A highly detailed concept may therefore remain weakly evidenced. Conversely, a simple mechanism may have strong empirical evidence. + +This distinction prevents sophistication of presentation from being mistaken for certainty. + +--- + +# 10. Recommended metadata + +C3 documents SHOULD include concise frontmatter. + +## Compendium + +```yaml +--- +title: "..." +artifact_type: compendium +status: active +last_reviewed: YYYY-MM-DD +evidence_level: E0-E6 +--- +``` + +## Catalog + +```yaml +--- +title: "..." +artifact_type: catalog +status: active +schema_version: "0.1" +last_updated: YYYY-MM-DD +--- +``` + +## Concept + +```yaml +--- +title: "..." +artifact_type: concept +status: active +maturity: exploratory +evidence_level: E1 +last_reviewed: YYYY-MM-DD +--- +``` + +Metadata should remain lightweight. C3 is intended to improve thinking, not create administrative overhead. + +--- + +# 11. Naming guidance + +C3 does not mandate one filename convention, but filenames SHOULD reveal artifact purpose. + +### Compendium + +Use `...Compendium.md` or a plain subject name where directory placement is sufficient. + +### Catalog + +Use `...Catalog.md`. + +Examples: + +```text +catalog/ +├── BenchmarkCatalog.md +├── MechanismCatalog.md +└── IndicatorCatalog.md +``` + +### Concepts + +Concepts SHOULD normally use the actual mechanism or design name. + +Examples: + +```text +concepts/ +├── CustomerEarnedEquityDesign.md +├── AdaptiveTransitionBuffer.md +└── IndividualParticipationCapitalAccount.md +``` + +A suffix such as `Design`, `Architecture`, `Mechanism`, or `Model` may be useful where ambiguity exists. + +--- + +# 12. Cross-linking between C3 artifacts + +C3 artifacts SHOULD reference related artifacts from other classes when those relationships are meaningful. + +Example: + +```yaml +related_compendium: + - ../compendium/ConsumerOwnership.md + +related_catalogs: + - ../catalog/ConsumerCapitalMechanismCatalog.md + +related_concepts: + - ./IndividualParticipationCapitalAccount.md +``` + +Narrative cross-links are equally acceptable. + +The objective is to make the knowledge graph visible without requiring a dedicated graph database. + +--- + +# 13. Example: Customer-Earned Equity + +The Customer-Earned Equity topic illustrates C3 well. + +## Compendium + +```text +compendium/ +└── ConsumerOwnership.md +``` + +Answers: + +- What forms of consumer ownership already exist? +- What does research say? +- What economic theory applies? +- What effects are empirically supported? + +## Catalog + +```text +catalog/ +└── ConsumerCapitalMechanismCatalog.md +``` + +Enumerates patronage dividends, mutual ownership, customer shares, loyalty-to-equity programmes, profit participation, warrants and diversified customer funds. + +## Concept + +```text +concepts/ +└── CustomerEarnedEquityDesign.md +``` + +Designs eligibility, accrual rules, capitalization, dilution, matching, diversification, portability, rights, tax treatment and failure controls. + +--- + +# 14. Anti-patterns + +## 14.1 The Everything Compendium + +Do not put every observation, source, idea and design proposal into the compendium. The compendium should represent synthesized knowledge. + +## 14.2 The Narrative Catalog + +Do not let catalog entries evolve into essays. If extensive interpretation is needed, move it into the compendium. + +## 14.3 The Concept Presented as Fact + +A concept SHOULD clearly state assumptions, maturity and evidence. A polished design is still a design. + +## 14.4 The Research Note That Becomes Canon by Age + +Old research notes do not automatically become authoritative because nobody revisited them. Durable conclusions should be incorporated into the compendium. + +## 14.5 Duplicate Truth + +Avoid maintaining the same canonical explanation in multiple places. + +Prefer: + +```text +compendium → canonical explanation +catalog → canonical enumeration +concept → canonical proposed design +``` + +Other artifacts should link to these. + +--- + +# 15. Lifecycle of a research result + +A typical result may evolve as follows. + +### Stage 1 — Observation + +Captured in: + +```text +research/ +``` + +### Stage 2 — Repeated evidence + +Structured in: + +```text +catalog/ +``` + +or synthesized into: + +```text +compendium/ +``` + +### Stage 3 — Design implication + +Developed in: + +```text +concepts/ +``` + +### Stage 4 — Experiment + +Recorded as research/data. + +### Stage 5 — Validation or falsification + +Updates `compendium/`, `catalog/`, and/or `concepts/` as appropriate. + +The repository therefore accumulates knowledge without erasing its developmental history. + +--- + +# 16. C3 and research governance + +C3 supports three complementary forms of governance. + +## Knowledge governance + +The compendium provides the current synthesized view. + +## Coverage governance + +The catalog reveals what has and has not been systematically identified. + +## Innovation governance + +The concepts directory exposes speculative ideas explicitly, making them reviewable without confusing them with established knowledge. + +This separation helps reviewers ask better questions: + +- **Compendium:** Is this conclusion supported? +- **Catalog:** Is this inventory complete and consistently classified? +- **Concept:** Could this mechanism work, and how would we know? + +--- + +# 17. C3 as an epistemic safety mechanism + +C3 is deliberately more than directory organization. + +It reduces several recurring research errors: + +- speculation becoming fact through repetition; +- catalogs being mistaken for explanation; +- source volume being mistaken for knowledge; +- mature prose being mistaken for mature evidence; +- proposed interventions being evaluated as if already validated; +- experimental outcomes failing to update canonical understanding. + +C3 therefore acts as a lightweight **epistemic type system**. + +In programming terms: + +```text +Compendium : KnowledgeModel +Catalog : StructuredInventory +Concept : ProposedIntervention +``` + +Mixing those types should be treated as a design smell. + +--- + +# 18. Minimal C3 repository + +The minimum compliant structure is: + +```text +research-project/ +├── catalog/ +├── compendium/ +├── concepts/ +└── README.md +``` + +--- + +# 19. Recommended research repository + +For continuously operated research systems such as `intelligence-radar`: + +```text +intelligence-radar/ +├── catalog/ +│ ├── benchmarks/ +│ ├── indicators/ +│ ├── mechanisms/ +│ └── sources/ +│ +├── compendium/ +│ ├── intelligence/ +│ ├── economics/ +│ ├── technology/ +│ └── policy/ +│ +├── concepts/ +│ ├── economics/ +│ ├── infrastructure/ +│ └── governance/ +│ +├── data/ +│ ├── raw/ +│ ├── normalized/ +│ └── derived/ +│ +├── digests/ +│ ├── daily/ +│ └── weekly/ +│ +├── research/ +│ └── YYYY/ +│ └── YYYY-MM-DD-*.md +│ +└── README.md +``` + +Subdirectories SHOULD emerge only when volume justifies them. The three top-level C directories remain the stable semantic anchors. + +--- + +# 20. C3 placement lint + +Before accepting a new durable artifact, apply this quick check. + +### Compendium lint + +- Does it synthesize rather than merely list? +- Does it distinguish evidence from uncertainty? +- Is it describing current knowledge rather than proposing a design? + +### Catalog lint + +- Does it enumerate identifiable things? +- Are entries structurally comparable? +- Can new entries be added without rewriting the whole artifact? + +### Concept lint + +- Is it proposing something? +- Are assumptions explicit? +- Are mechanism and actors clear? +- Are failure modes considered? +- Is maturity/evidence stated? + +If the artifact fails all three tests, it probably belongs in `research/`, `data/`, or `digests/`. + +--- + +# 21. Standard summary + +The **C3ResearchDeliverablesStandard** organizes durable research output into three semantic classes: + +| Class | Core question | Function | +|---|---|---| +| **Compendium** | What do we know? | Synthesize knowledge | +| **Catalog** | What have we identified? | Structure inventories | +| **Concepts** | What could we create or change? | Design possibilities | + +Supporting material remains outside the C3 taxonomy: + +| Directory | Function | +|---|---| +| `research/` | Investigations and working research | +| `digests/` | Time-bounded change summaries | +| `data/` | Raw and structured evidence | + +The fundamental rule is: + +> **Compendium = models of the world.** +> **Catalog = inventory of the world.** +> **Concepts = possible changes to the world.** + +Operationally: + +> **Understand → Invent → Enumerate → Test → Learn.** + +C3 is intentionally small. + +Its value comes from maintaining the distinction consistently. diff --git a/concepts/CustomerEarnedEquityDesign.md b/concepts/CustomerEarnedEquityDesign.md new file mode 100755 index 0000000..bc5020f --- /dev/null +++ b/concepts/CustomerEarnedEquityDesign.md @@ -0,0 +1,1716 @@ +--- +title: "Customer-Earned Equity Design" +short_title: "CustomerEarnedEquityDesign" +version: "0.1" +status: "exploratory" +artifact_type: "concept" +maturity: "exploratory" +evidence_level: "E1" +project: + - intelligent-economics + - intelligence-radar + - coulomb +related_concepts: + - ConsumerCapitalParticipation + - IndividualParticipationCapitalAccount + - AdaptiveTransitionBuffer +tags: + - customer-equity + - consumer-capital + - ownership + - participation + - abundance-economics + - progressive-capital-formation +--- + +# Customer-Earned Equity Design + +**Version:** 0.1 +**Artifact type:** Concept +**Maturity:** Exploratory +**Evidence level:** E1 — historical analogues exist; the integrated mechanism proposed here is not yet validated. + +--- + +# 1. Purpose + +**Customer-Earned Equity (CEE)** is a proposed mechanism through which a customer can accumulate a productive capital claim in a company because of a sustained economic relationship with that company. + +The concept responds to a possible structural consequence of AI-driven abundance: + +> production becomes increasingly scalable while human labour becomes less central to the generation of economic output. + +If this occurs, wages may become a weaker mechanism for distributing purchasing power and productive ownership. + +CEE explores a complementary participation mechanism: + +> **customers who help a company learn, survive, grow and scale can accumulate a durable claim on some of that scalability.** + +The concept is not intended to replace: + +- wages; +- entrepreneurship; +- investment; +- competitive markets; +- social insurance; +- ordinary consumer choice. + +It adds another route through which individuals may accumulate productive capital. + +--- + +# 2. Core proposition + +A customer may contribute more to enterprise value than the purchase price alone represents. + +Depending on context, the customer can provide: + +- demand; +- attention; +- trust; +- early adoption; +- recurring revenue; +- product feedback; +- usage data explicitly volunteered for improvement; +- referrals; +- validation; +- tolerance for product immaturity; +- participation in experimentation; +- reputation transfer; +- long-term market signal. + +In highly scalable markets, these inputs can become relatively more valuable because production itself is less scarce than: + +- identifying useful demand; +- earning trust; +- selecting among abundant alternatives; +- learning what people actually value. + +CEE therefore asks: + +> **Should part of the value generated through a durable customer relationship be converted into productive ownership rather than only immediate consumption rewards?** + +--- + +# 3. Design objectives + +CEE should attempt to satisfy the following objectives simultaneously. + +## 3.1 Capital formation + +Create individually owned productive assets. + +## 3.2 Agency + +Make the individual a participant in economic upside rather than only a recipient of discounts or transfers. + +## 3.3 Demand alignment + +Reward customers whose sustained participation contributes to product-market fit and enterprise learning. + +## 3.4 Long-term orientation + +Encourage durable value creation rather than purely transactional consumption. + +## 3.5 Competition preservation + +Avoid creating customer lock-in that weakens market discipline. + +## 3.6 Progressivity + +Avoid a structure in which wealthy customers automatically accumulate disproportionate ownership simply because they can spend more. + +## 3.7 Simplicity + +Make the mechanism understandable enough to be used by ordinary consumers. + +## 3.8 Privacy + +Do not make surveillance a prerequisite for participation. + +## 3.9 Portability + +Ownership should survive the end of the customer relationship. + +## 3.10 Falsifiability + +CEE should be testable against alternative reward mechanisms such as: + +- cashback; +- discounts; +- loyalty points; +- profit sharing; +- broad-market investment contributions; +- unconditional cash. + +--- + +# 4. Non-goals + +CEE is not intended to: + +- preserve a firm's customer base artificially; +- guarantee investment returns; +- compensate customers for every purchase; +- turn every customer into a voting shareholder; +- replace consumer protection; +- prevent company failure; +- eliminate wealth inequality; +- socialize ownership through compulsory collective structures; +- reward addictive engagement; +- create speculative loyalty tokens. + +CEE should remain a **capital formation mechanism attached to real economic participation**. + +--- + +# 5. Actors + +The minimum system contains four actors. + +## 5.1 Customer + +The individual who: + +- purchases; +- uses; +- contributes; +- remains in the relationship; +- earns a capital claim. + +## 5.2 Provider + +The company whose economic activity generates the underlying asset or profit claim. + +## 5.3 Custodian / participation account provider + +Holds or records fractional claims on behalf of the customer. + +This could eventually be an **Individual Participation Capital Account (IPCA)** provider. + +## 5.4 Public policy layer + +Optional. + +May provide: + +- tax treatment; +- progressive matching; +- regulation; +- investor protection; +- anti-fraud rules; +- portability rules. + +The concept should remain viable even without public matching. + +--- + +# 6. Basic mechanism + +A minimal CEE mechanism looks like this: + +```text +customer relationship + ↓ +qualifying participation + ↓ +equity credit accrual + ↓ +periodic conversion + ↓ +capital instrument + ↓ +individual ownership +``` + +The customer earns an **Equity Credit**. + +An Equity Credit is not itself necessarily a security. + +It is an accounting unit that determines how much qualified capital the customer receives at a defined conversion event. + +Example: + +```text +€1,000 qualifying annual customer value +× 2% CEE rate += €20 customer equity credit +``` + +At conversion: + +```text +€20 equity credit +→ €20 of qualifying company equity +``` + +or: + +```text +€20 equity credit +→ €20 contribution to diversified participation fund +``` + +depending on the chosen architecture. + +--- + +# 7. What should generate CEE? + +CEE should not automatically equal spending. + +A better model is a weighted contribution function. + +A first generic formulation: + +```text +CEE Score = + Spend Contribution + + Tenure Contribution + + Early-Adopter Contribution + + Product Contribution + + Referral Contribution + + Community Contribution +``` + +Each component should be capped. + +This prevents any single dimension from dominating. + +--- + +# 8. Spend contribution + +Spending is the easiest variable to measure and the most obvious economic contribution. + +However, it is also the most regressive. + +A simple first design could use diminishing returns. + +Example: + +```text +first €500 annual spend: 2.0% equity credit +next €1,500: 1.0% +next €3,000: 0.5% +above €5,000: 0.1% +``` + +This gives: + +- strong participation for normal customers; +- diminishing advantage for very high spenders; +- a natural cap on concentration. + +Alternative: + +```text +CEE spend credit = k × sqrt(annual spend) +``` + +This makes accrual sublinear rather than proportional. + +--- + +# 9. Tenure contribution + +Long-term customers provide: + +- recurring demand; +- lower acquisition cost; +- better revenue predictability; +- stronger product feedback over time. + +CEE can therefore reward relationship duration. + +Example: + +```text +Year 1: 1.00× multiplier +Year 2: 1.05× +Year 3: 1.10× +Year 4: 1.15× +Year 5+: 1.20× cap +``` + +The multiplier should remain modest. + +Otherwise tenure becomes lock-in. + +--- + +# 10. Early-adopter contribution + +Early customers may create disproportionate value because they: + +- accept higher product risk; +- provide formative feedback; +- validate a market; +- create reference credibility. + +A company could reserve a fixed **Early Customer Pool**. + +Example: + +```text +0.50% of fully diluted equity +reserved for first 10,000 qualifying customers +``` + +Allocation could depend on: + +- adoption order; +- active usage; +- minimum relationship duration; +- contribution to product feedback. + +This should be transparent before customers participate. + +Retroactive rule changes should be prohibited or tightly constrained. + +--- + +# 11. Product contribution + +Customers may voluntarily provide high-value input beyond purchases. + +Examples: + +- structured product feedback; +- bug reports; +- usability testing; +- beta participation; +- documented workflows; +- design interviews; +- verified feature suggestions; +- community support. + +The mechanism should reward **useful contribution**, not raw engagement volume. + +Bad metric: + +```text +number of clicks +``` + +Better metric: + +```text +accepted improvement contribution +``` + +or: + +```text +completed structured research participation +``` + +Product contribution should be: + +- explicit; +- transparent; +- privacy preserving; +- independently auditable where economically meaningful. + +--- + +# 12. Referral contribution + +Customer acquisition is economically valuable. + +A referral contribution may therefore be appropriate. + +Possible structure: + +```text +qualified referred customer +→ fixed equity credit +``` + +rather than: + +```text +percentage of referred customer's future spend +``` + +The latter risks creating multi-level incentive structures. + +CEE should avoid: + +- pyramid incentives; +- recursive commissions; +- speculative recruitment. + +--- + +# 13. Community contribution + +Some products gain value from: + +- peer support; +- tutorials; +- templates; +- curation; +- moderation; +- knowledge contributions. + +These contributions may justify CEE when they have measurable productive value. + +But community contribution is also one of the easiest areas to game. + +It should therefore be: + +- capped; +- quality weighted; +- explicitly reviewed; +- optional; +- not based on surveillance-derived engagement scores. + +--- + +# 14. Accrual formula + +A first generic CEE formula: + +```text +CEE Credit += +Base Customer Credit +× Contribution Score +× Tenure Multiplier +``` + +Where: + +```text +Contribution Score = + ws × SpendScore ++ wp × ProductContribution ++ wr × ReferralContribution ++ wc × CommunityContribution ++ we × EarlyAdopterContribution +``` + +with: + +```text +ws + wp + wr + wc + we = 1 +``` + +Every term should have: + +- a public definition; +- a cap; +- a revision policy; +- a dispute process. + +--- + +# 15. Equity source + +CEE must answer a fundamental question: + +> **Where does the equity come from?** + +Possible architectures follow. + +--- + +# 16. Architecture A — New equity issuance + +The company issues new shares to satisfy CEE claims. + +Advantages: + +- simple economic logic; +- preserves company cash; +- directly connects customer participation to ownership. + +Disadvantages: + +- shareholder dilution; +- cap-table complexity; +- potential investor resistance. + +Suitable for: + +- startups; +- growth companies; +- firms that deliberately reserve a CEE pool. + +--- + +# 17. Architecture B — Treasury shares + +The company distributes shares already held in treasury. + +Advantages: + +- no immediate new issuance; +- administratively cleaner where treasury stock exists. + +Disadvantages: + +- limited supply; +- still transfers value from existing shareholders. + +--- + +# 18. Architecture C — Market purchase + +The provider pays cash into a customer participation account that purchases company shares in the market. + +Advantages: + +- avoids dilution; +- works for listed firms; +- price is externally determined. + +Disadvantages: + +- cash cost; +- unavailable for private firms; +- can become expensive during high valuations. + +--- + +# 19. Architecture D — Profit participation unit + +Instead of common equity, customers receive an economic claim on: + +- profits; +- revenues; +- dividends; +- defined distributable surplus. + +Advantages: + +- simpler governance; +- no voting complexity; +- easier to tailor. + +Disadvantages: + +- potentially weaker capital appreciation; +- legal treatment may still be securities-like; +- claim design can become opaque. + +--- + +# 20. Architecture E — Participation fund contribution + +The provider contributes value into a diversified fund rather than issuing direct firm equity. + +Advantages: + +- diversification; +- reduced lock-in; +- simpler individual risk profile. + +Disadvantages: + +- weaker psychological connection to the provider; +- less direct alignment; +- requires fund infrastructure. + +--- + +# 21. Architecture F — Hybrid + +A hybrid may be preferable. + +Example: + +```text +50% company-specific equity +50% diversified participation fund +``` + +This preserves: + +- connection; +- ownership meaning; +- diversification. + +A default hybrid should be tested against direct-equity and fully diversified alternatives. + +--- + +# 22. Recommended initial architecture + +For early experimentation: + +> **Provider-funded capital contribution + capped provider-specific exposure + default diversification** + +Example: + +```text +CEE earned: €100 + +€40 → provider-specific equity +€60 → diversified participation fund +``` + +This reduces: + +- customer lock-in; +- single-firm risk; +- correlated exposure. + +The ratio should remain experimental. + +--- + +# 23. Equity pool design + +A company adopting CEE should define a maximum economic budget. + +Example: + +```text +Customer Participation Pool: +2% of fully diluted equity over 5 years +``` + +Possible sub-pools: + +```text +0.5% Early Customer Pool +0.8% Ongoing Customer Participation +0.4% Product Contribution +0.3% Referral / Community Contribution +``` + +The pool should be: + +- disclosed; +- capped; +- predictable; +- included in financing models. + +The CEE mechanism should not create uncontrolled dilution. + +--- + +# 24. Progressive matching + +CEE becomes more interesting when paired with **progressive capital matching**. + +The company creates the base claim. + +Public policy may amplify that claim for individuals with low income or low existing productive capital. + +Example: + +```text +company CEE earned: €100 +public match: €150 +total capital: €250 +``` + +A high-income participant might receive: + +```text +company CEE earned: €100 +public match: €10 +total capital: €110 +``` + +--- + +# 25. Matching basis + +Matching should preferably depend on: + +- taxable income; +- household capital ownership; +- or another simple observable proxy. + +Avoid: + +- complex behavioural scoring; +- discretionary caseworker judgement; +- intrusive financial profiling. + +Possible formula: + +```text +match_rate = +min(max_match, + base_rate × income_adjustment × wealth_adjustment) +``` + +A simpler policy may be better: + +```text +bottom income quartile: 200% +second quartile: 100% +third quartile: 50% +top quartile: 0% +``` + +This should be evaluated for: + +- cliff effects; +- gaming; +- administration; +- fairness. + +--- + +# 26. Matching cap + +Public matching should be capped per person. + +Example: + +```text +maximum annual qualifying match: +€1,000 per individual +``` + +This limits fiscal exposure and reduces incentives for artificial consumption. + +--- + +# 27. Do not reward consumption for its own sake + +A serious failure mode would be: + +> buy more products in order to receive subsidized equity. + +The public matching mechanism should therefore avoid acting as a consumption subsidy. + +Possible controls: + +- cap spend-derived CEE; +- weight tenure and contribution; +- match only CEE generated within normal purchasing patterns; +- apply annual person-level caps across providers; +- exclude financed or circular transactions. + +--- + +# 28. Portability + +CEE should create a real asset. + +It must therefore survive: + +- customer churn; +- provider switching; +- dissatisfaction; +- relocation; +- changes in consumption. + +A customer who leaves the provider keeps vested CEE. + +This is essential. + +Without portability, CEE becomes a loyalty trap rather than ownership. + +--- + +# 29. Vesting + +Some delay between earning and final ownership may be justified to prevent gaming. + +However, long vesting can undermine consumer rights. + +A reasonable first structure: + +```text +monthly accrual +quarterly vesting +``` + +or: + +```text +accrual during period +vesting after 90 days +``` + +Vesting should not be contingent on continued purchase after the earning period. + +--- + +# 30. Liquidity + +Once vested, the customer should have clear options. + +For public-company equity: + +- hold; +- sell; +- diversify. + +For private-company equity: + +- hold; +- transfer to secondary market if legally possible; +- convert to participation fund under predefined rules; +- participate in company liquidity events. + +A CEE design should not pretend that illiquid private equity equals cash. + +Valuation and liquidity risk must be explicit. + +--- + +# 31. Customer rights + +CEE can separate **economic rights** from **control rights**. + +Possible classes: + +## Economic-only + +- dividend rights; +- liquidation rights; +- capital appreciation; +- no vote. + +## Limited governance + +- customer advisory votes; +- defined class votes; +- representation rights. + +## Full shareholder rights + +- standard voting shares. + +The default should likely begin with **economic ownership plus limited customer-specific governance**, not full governance rights. + +The mechanism should be tested rather than assumed. + +--- + +# 32. Customer voice + +One useful addition is a distinct **Customer Participation Assembly** or digital advisory process. + +Its purpose would not be to manage the company. + +It could provide structured input on: + +- customer-impacting policy; +- major product direction; +- data practices; +- pricing principles; +- CEE rule changes. + +The design should preserve managerial responsibility. + +--- + +# 33. Provider switching + +A good CEE system must preserve competition. + +A customer should be able to: + +```text +stop buying provider A +↓ +keep vested equity in A +↓ +begin earning equity in provider B +``` + +There should be no penalty for switching. + +This creates an interesting dynamic: + +> customers can stop consuming from a company while remaining owners. + +That may improve long-term market discipline. + +--- + +# 34. Diversification + +Direct provider ownership can create correlated risk: + +```text +provider fails +→ product disappears +→ customer loses service +→ customer equity also loses value +``` + +CEE should therefore offer: + +- direct equity caps; +- default diversification; +- automatic index conversion above a threshold; +- optional portfolio rebalancing. + +Example: + +```text +maximum provider-specific CEE exposure: +20% of total participation account +``` + +The rest could be redirected to a diversified participation fund. + +--- + +# 35. Interaction with the Individual Participation Capital Account + +CEE should ideally settle into a portable **Individual Participation Capital Account (IPCA)**. + +The IPCA could combine: + +```text +customer-earned equity ++ employee ownership ++ entrepreneurial ownership ++ matched contributions ++ voluntary investment +``` + +This prevents CEE from becoming a fragmented collection of tiny accounts. + +The IPCA becomes the individual's long-term productive capital layer. + +--- + +# 36. Relationship to ordinary loyalty programs + +CEE differs fundamentally from points. + +| Loyalty points | Customer-Earned Equity | +|---|---| +| liability of provider | productive capital claim | +| often expires | should vest permanently | +| usually spendable only with provider | portable asset | +| promotes future consumption | builds capital | +| no upside from provider growth | participates in upside | +| provider controls rules | rights should be legally defined | + +CEE could coexist with loyalty programs, but should not be presented as another points scheme. + +--- + +# 37. Relationship to cashback + +Cashback optimizes immediate customer value. + +CEE optimizes: + +- long-term ownership; +- compounding; +- alignment; +- productive participation. + +The research programme should directly compare: + +```text +€100 cashback +vs +€100 provider equity +vs +€100 diversified equity +vs +€100 matched participation capital +``` + +--- + +# 38. Relationship to UBI + +CEE is not an income floor. + +It is not appropriate as a substitute for emergency support. + +A person facing immediate material hardship cannot necessarily wait for equity to compound. + +The relevant contrast is therefore not: + +```text +CEE OR UBI +``` + +but: + +```text +income stabilization ++ +capital participation +``` + +The research question is whether a system emphasizing capital formation produces stronger long-term: + +- agency; +- wealth; +- resilience; +- optionality; +- economic participation. + +--- + +# 39. Psychological design + +CEE should avoid messaging such as: + +> "You have been rewarded for being loyal." + +That frames the customer as a passive target. + +A stronger framing is: + +> **Your participation helped create value. Part of that value is now yours as productive capital.** + +The distinction matters. + +Desired psychological effects: + +- agency; +- future orientation; +- ownership; +- experimentation; +- financial literacy; +- sense of contribution. + +Potential negative effects: + +- sunk-cost attachment; +- irrational loyalty; +- overconsumption; +- speculative behaviour; +- identity fusion with brands. + +These effects must be measured. + +--- + +# 40. Pricing interaction + +CEE should be economically visible. + +A company offering CEE may choose among several strategies: + +## Strategy A — Same price, lower margin + +CEE is funded from margin. + +## Strategy B — Higher price, ownership included + +Customer explicitly purchases a bundle: + +```text +product + participation capital +``` + +## Strategy C — Marketing substitution + +CEE replaces part of: + +- advertising; +- discounts; +- loyalty rewards; +- referral commissions. + +This may be particularly attractive. + +Example: + +```text +current customer acquisition cost: €150 +CEE acquisition allocation: €75 +``` + +Instead of paying a platform for attention, the company gives part of acquisition economics directly to the customer. + +This may be one of the strongest business cases for CEE. + +--- + +# 41. Marketing-budget substitution hypothesis + +Many firms spend heavily on: + +- advertising platforms; +- affiliate fees; +- discounts; +- loyalty programs; +- referral bonuses. + +CEE could redirect part of this budget: + +```text +marketing intermediary + ↓ +customer capital +``` + +Hypothesis: + +> firms can acquire and retain customers more efficiently by giving customers productive ownership than by buying repeated access to those customers from intermediaries. + +This deserves direct experimental testing. + +--- + +# 42. Company valuation implications + +CEE may affect valuation positively through: + +- lower churn; +- lower acquisition cost; +- higher customer lifetime value; +- stronger advocacy; +- better product feedback; +- community effects. + +Negative effects may include: + +- dilution; +- administration; +- investor uncertainty; +- governance complexity. + +The firm-level economics must therefore be modeled as: + +```text +CEE Net Value += +Retention Gain ++ Acquisition Savings ++ Referral Gain ++ Product Learning Gain +- Dilution Cost +- Administration Cost +- Governance Cost +- Lock-in / Distortion Cost +``` + +--- + +# 43. Anti-gaming controls + +CEE systems are highly gameable if poorly designed. + +Potential attacks: + +- artificial purchases; +- wash transactions; +- fake referrals; +- contribution spam; +- account farming; +- synthetic identities; +- coordinated rating manipulation. + +Controls may include: + +- economic minimums; +- delayed vesting; +- contribution review; +- person-level caps; +- fraud detection; +- transaction finality; +- reversal rules for refunds; +- identity uniqueness where legally justified. + +Controls should remain proportional. + +CEE should not require intrusive identity surveillance by default. + +--- + +# 44. Refunds and reversals + +If a purchase is refunded: + +```text +unvested spend-derived CEE +→ reversed +``` + +Vested CEE should generally not be clawed back except in fraud cases. + +This gives ownership legal finality. + +--- + +# 45. Failure and insolvency + +CEE is real capital and can lose value. + +The customer must understand: + +```text +company fails +→ equity may become worthless +``` + +This is not a bug. + +Capital participation includes risk. + +The design goal is not to remove risk, but to: + +- cap concentration; +- diversify; +- communicate clearly; +- avoid presenting equity as guaranteed reward value. + +--- + +# 46. Acquisition of the provider + +If the provider is acquired: + +CEE holders should participate according to the economic rights of their instrument. + +Possible outcome: + +```text +customer equity +→ cash +``` + +or: + +```text +customer equity +→ acquiring-company equity +``` + +The treatment should be defined before issuance. + +--- + +# 47. Customer death + +Vested CEE should normally be inheritable like other property. + +This reinforces its status as genuine capital. + +Alternative policy treatments should be evaluated separately. + +--- + +# 48. Minors + +CEE for minors raises additional issues. + +Possible architecture: + +```text +minor earns CEE +→ held in custodial participation account +→ control transfers at legal adulthood +``` + +This could become interesting for long-term consumer-capital formation, but requires careful legal treatment. + +--- + +# 49. Essential services + +CEE may be particularly sensitive in markets such as: + +- utilities; +- healthcare; +- insurance; +- housing; +- education. + +Customer choice may already be constrained. + +Therefore: + +> CEE should never substitute for consumer rights or create pressure to remain with a provider of essential services. + +Direct-equity exposure may also be inappropriate where service dependency is already high. + +--- + +# 50. Privacy model + +CEE should follow: + +> **contribution transparency without behavioral surveillance.** + +Preferred data: + +- purchases; +- explicit tenure; +- explicit referrals; +- explicit participation; +- explicit accepted contributions. + +Avoid: + +- hidden engagement scoring; +- cross-site tracking; +- inferred personality; +- emotional manipulation; +- attention extraction. + +--- + +# 51. Explainability + +Every customer should be able to inspect: + +```text +CEE earned this period: €X + +from purchases: €A +from tenure: €B +from contribution: €C +from referral: €D +public match: €E +``` + +A mechanism that cannot explain an individual's accrual should not be used. + +--- + +# 52. Rule changes + +Providers may need to change the programme. + +However: + +- vested claims must not be altered; +- accrued but unvested claims should be protected where feasible; +- prospective rule changes require notice; +- material changes should be documented. + +CEE rules should behave more like financial rights than gamified loyalty terms. + +--- + +# 53. Standardization opportunity + +A future CEE standard could define: + +- accrual disclosure; +- vesting rules; +- portability; +- valuation; +- rights; +- diversification; +- tax reporting; +- fraud controls; +- rule-change policy. + +A standard may improve trust and comparability. + +Possible future artifact: + +```text +CustomerEarnedEquityProtocol.md +``` + +--- + +# 54. Pilot design + +A first pilot should be small, voluntary and measurable. + +Recommended target: + +- startup or small subscription company; +- recurring customer relationship; +- measurable retention; +- low regulatory complexity; +- customer base willing to experiment. + +Avoid first pilots in: + +- essential services; +- highly regulated finance; +- healthcare; +- high-risk speculative markets. + +--- + +# 55. Pilot structure + +Example: + +```text +Pilot duration: 12 months +Eligible customers: 1,000 +CEE pool: €25,000 equivalent +Direct provider share: 40% +Diversified share: 60% +Annual per-user cap: €250 +Vesting: quarterly +``` + +Randomized groups: + +```text +A — conventional loyalty points +B — cashback +C — provider equity +D — hybrid CEE +``` + +Measure: + +- retention; +- referrals; +- average spend; +- product feedback; +- customer understanding; +- perceived agency; +- perceived fairness; +- switching willingness; +- trust; +- investment behaviour. + +--- + +# 56. Progressive-match pilot + +A separate pilot could compare: + +```text +CEE only +CEE + 50% match +CEE + 100% match +CEE + income-sensitive match +``` +CustomerEarnedEquityDesign.md is cle +Measure: + +- wealth formation; +- consumption distortion; +- perceived fairness; +- agency; +- participation by income bracket. + +--- + +# 57. Key hypotheses to test + +## H1 + +CEE produces higher perceived long-term value than equivalent cashback. + +## H2 + +CEE improves retention without materially reducing willingness to switch when provider quality declines. + +## H3 + +Hybrid direct + diversified CEE produces better risk-adjusted outcomes than direct equity alone. + +## H4 + +Progressive matching increases productive capital formation among low-capital participants without materially increasing wasteful consumption. + +## H5 + +CEE improves product feedback and customer advocacy. + +## H6 + +CEE can substitute for part of customer acquisition and loyalty expenditure. + +## H7 + +CEE strengthens perceived economic agency more than equivalent loyalty rewards. + +## H8 + +Very small equity amounts do not create meaningful behavioural effects below a minimum threshold. + +--- + +# 58. Critical falsifiers + +CEE should be rejected or heavily revised if evidence shows that it systematically: + +- encourages overconsumption; +- produces strong lock-in; +- worsens competition; +- creates negligible wealth; +- mainly benefits high-income customers; +- imposes excessive administration; +- causes customers to hold dangerously concentrated assets; +- fails to outperform simpler cashback; +- becomes primarily a marketing gimmick; +- encourages surveillance; +- weakens customer switching; +- creates securities complexity disproportionate to value. + +--- + +# 59. Minimum viable CEE + +The smallest meaningful implementation may be: + +```text +1. customer makes qualifying purchase +2. provider allocates fixed % of relationship value +3. amount settles quarterly +4. customer receives portable capital asset +5. vested asset survives churn +6. provider-specific exposure is capped +7. customer can diversify +8. accrual is transparent +``` + +Everything else is optional enhancement. + +--- + +# 60. Preferred first principles + +CEE v0.1 adopts the following first principles: + +1. **Ownership must be real.** +2. **Vested claims must be portable.** +3. **Rules must be explainable.** +4. **Spend must not dominate accrual.** +5. **High spending must face diminishing returns.** +6. **Participation should be capped.** +7. **Diversification should be the default.** +8. **Customer switching must remain easy.** +9. **Public matching should build capital, not subsidize consumption.** +10. **Privacy should be preserved.** +11. **No speculative token is required.** +12. **The mechanism must outperform simpler alternatives to justify its complexity.** + +--- + +# 61. Open design questions + +1. What is the correct economic basis for allocating CEE? +2. Should spend, tenure or contribution dominate? +3. What minimum amount makes CEE psychologically meaningful? +4. What fraction should be direct equity versus diversified capital? +5. Should customer equity carry voting rights? +6. How should private-company shares be valued? +7. How can fractional private equity be administered cheaply? +8. What should happen when a company changes legal form? +9. How should public matching be capped? +10. Should matching depend on income, wealth, or both? +11. How can tax policy avoid subsidizing consumption? +12. Can CEE materially lower customer acquisition cost? +13. Does CEE improve innovation through higher-quality feedback? +14. Does CEE create anti-competitive loyalty? +15. Should customers be allowed to immediately sell? +16. Would minimum holding periods increase long-term orientation or merely reduce agency? +17. Should CEE be inheritable? +18. Should CEE be protected from creditors? +19. How should minors participate? +20. What legal form best supports pan-European CEE? + +--- + +# 62. Research dependencies + +CEE depends on several other research areas. + +## Consumer ownership history + +Needed to understand existing mechanisms and failures. + +## Shared capitalism + +Needed to compare ownership + participation effects. + +## Basic income / transfer evidence + +Needed to compare agency and welfare effects. + +## Wealth-building accounts + +Needed to understand matching and long-term capital formation. + +## Securities law + +Needed to make fractional claims feasible. + +## Behavioural economics + +Needed to understand lock-in, agency and risk perception. + +## Market design + +Needed to preserve switching and competition. + +--- + +# 63. Relationship to C3 deliverables + +This file belongs in: + +```text +concepts/ +``` + +because it proposes a mechanism. + +Related future artifacts: + +```text +compendium/ +└── ConsumerOwnership.md + +catalog/ +├── ConsumerCapitalMechanismCatalog.md +└── CustomerOwnershipPrecedentCatalog.md + +research/ +├── CustomerEquityEvidenceReview.md +└── CustomerEquityLegalResearch.md +``` + +Evidence from those artifacts should update this concept. + +--- + +# 64. Version 0.1 assessment + +Customer-Earned Equity appears conceptually plausible and has partial historical precedents. + +Its strongest strategic logic is: + +> **redirect part of the economics currently spent on acquisition, retention and loyalty into durable customer-owned productive capital.** + +Its strongest macroeconomic logic is: + +> **create an additional path from consumer participation to capital ownership when labour becomes a weaker distribution mechanism.** + +Its strongest psychological claim is: + +> **ownership may preserve or increase agency better than purely passive reward systems.** + +Its principal risks are: + +- regressivity; +- loyalty lock-in; +- securities complexity; +- correlated risk; +- surveillance; +- trivial ownership; +- administrative cost. + +The concept is therefore sufficiently promising to justify: + +1. mechanism modeling; +2. legal feasibility research; +3. behavioral experiments; +4. a small real-world pilot. + +--- + +# 65. Working one-liner + +> **Customer-Earned Equity turns durable customer participation into portable productive capital.** + +Alternative formulation: + +> **When customers help create scalable value, part of the scalability should be able to become theirs.** diff --git a/data/artifacts/3e331ed3352d4132bef85b2f3d46cb1faeef612cc6d5ae1db68c36ec9145b9f9 b/data/artifacts/3e331ed3352d4132bef85b2f3d46cb1faeef612cc6d5ae1db68c36ec9145b9f9 new file mode 100644 index 0000000..a5f2636 --- /dev/null +++ b/data/artifacts/3e331ed3352d4132bef85b2f3d46cb1faeef612cc6d5ae1db68c36ec9145b9f9 @@ -0,0 +1 @@ +{"data":[{"id":"~deepseek/deepseek-pro-latest","canonical_slug":"~deepseek/deepseek-pro-latest","alias_target":{"name":"DeepSeek: DeepSeek V4 Pro 0813","slug":"deepseek/deepseek-v4-pro-0813"},"hugging_face_id":null,"name":"DeepSeek: DeepSeek Pro Latest","created":1789399174,"description":"This model always redirects to the latest model in the DeepSeek Pro family.","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.00000057948","completion":"0.00000173844","input_cache_read":"0.000000018438"},"top_provider":{"context_length":1048576,"max_completion_tokens":393216,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":1},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~deepseek/deepseek-pro-latest/endpoints"},"reasoning":{"mandatory":false,"supported_efforts":["max","high","low"],"default_effort":"high"}},{"id":"~deepseek/deepseek-flash-latest","canonical_slug":"~deepseek/deepseek-flash-latest","alias_target":{"name":"DeepSeek: DeepSeek V4.1 Flash","slug":"deepseek/deepseek-v4.1-flash"},"hugging_face_id":null,"name":"DeepSeek: DeepSeek Flash Latest","created":1789399150,"description":"This model always redirects to the latest model in the DeepSeek Flash family.","context_length":1048576,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000006","input_cache_read":"0.000000003"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~deepseek/deepseek-flash-latest/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"high"}},{"id":"inference-net/schematron-v2-turbo","canonical_slug":"inference-net/schematron-v2-turbo-20260902","hugging_face_id":"inference-net/schematron-v2-granite-4.0-h-micro","name":"Inference.net: Schematron V2 Turbo","created":1789176949,"description":"Schematron V2 Turbo is a 3B-parameter HTML-to-JSON extraction model from Inference.net. It prioritizes throughput for high-volume extraction workloads. Extraction instructions must be supplied through a JSON schema in response_format rather...","context_length":128000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000003","completion":"0.00000015","input_cache_read":"0.00000003"},"top_provider":{"context_length":128000,"max_completion_tokens":8192,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/inference-net/schematron-v2-turbo-20260902/endpoints"}},{"id":"inference-net/schematron-v2-small","canonical_slug":"inference-net/schematron-v2-small-20260902","hugging_face_id":"inference-net/schematron-v2-llama-3.2-3b","name":"Inference.net: Schematron V2 Small","created":1789176933,"description":"Schematron V2 Small is a 3B-parameter HTML-to-JSON extraction model from Inference.net. It prioritizes extraction quality for complex schemas and long pages. Extraction instructions must be supplied through a JSON schema...","context_length":128000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000005","completion":"0.00000023","input_cache_read":"0.00000005"},"top_provider":{"context_length":128000,"max_completion_tokens":4096,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/inference-net/schematron-v2-small-20260902/endpoints"}},{"id":"~openai/gpt-astra-latest","canonical_slug":"~openai/gpt-astra-latest","alias_target":{"name":"OpenAI: GPT-6 Astra","slug":"openai/gpt-6-astra"},"hugging_face_id":null,"name":"OpenAI: GPT Astra Latest","created":1789130932,"description":"This model always redirects to the latest model in the GPT Astra family.","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.00001","completion":"0.00005","web_search":"0.01","input_cache_read":"0.000001","input_cache_write":"0.0000125","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00002","completion":"0.000075","input_cache_read":"0.000002","input_cache_write":"0.000025"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~openai/gpt-astra-latest/endpoints"},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"medium"}},{"id":"~openai/gpt-sol-latest","canonical_slug":"~openai/gpt-sol-latest","alias_target":{"name":"OpenAI: GPT-5.6 Sol","slug":"openai/gpt-5.6-sol"},"hugging_face_id":null,"name":"OpenAI: GPT Sol Latest","created":1789130928,"description":"This model always redirects to the latest model in the GPT Sol family.","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.00001","web_search":"0.01","input_cache_read":"0.0000002","input_cache_write":"0.0000025","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000004","completion":"0.000015","input_cache_read":"0.0000004","input_cache_write":"0.000005"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/~openai/gpt-sol-latest/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"~openai/gpt-terra-latest","canonical_slug":"~openai/gpt-terra-latest","alias_target":{"name":"OpenAI: GPT-5.6 Terra","slug":"openai/gpt-5.6-terra"},"hugging_face_id":null,"name":"OpenAI: GPT Terra Latest","created":1789130925,"description":"This model always redirects to the latest model in the GPT Terra family.","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000012","web_search":"0.01","input_cache_read":"0.0000002","input_cache_write":"0.0000025","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000004","completion":"0.000018","input_cache_read":"0.0000004","input_cache_write":"0.000005"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/~openai/gpt-terra-latest/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"~openai/gpt-luna-latest","canonical_slug":"~openai/gpt-luna-latest","alias_target":{"name":"OpenAI: GPT-5.6 Luna","slug":"openai/gpt-5.6-luna"},"hugging_face_id":null,"name":"OpenAI: GPT Luna Latest","created":1789130922,"description":"This model always redirects to the latest model in the GPT Luna family.","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.0000012","web_search":"0.01","input_cache_read":"0.00000002","input_cache_write":"0.00000025","overrides":[{"min_prompt_tokens":272000,"prompt":"0.0000004","completion":"0.0000018","input_cache_read":"0.00000004","input_cache_write":"0.0000005"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/~openai/gpt-luna-latest/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"sakana/fugu-ultra-v2","canonical_slug":"sakana/fugu-ultra-v2-20260911","hugging_face_id":null,"name":"Sakana: Fugu Ultra v2","created":1789105383,"description":"Fugu Ultra v2 is the higher-performance model in Sakana AI's Fugu family. Rather than a single monolithic model, Fugu is a learned multi-agent orchestration system: a language model trained to...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.00003","web_search":"0.01","input_cache_read":"0.0000005","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00001","completion":"0.000045","input_cache_read":"0.000001"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","reasoning","reasoning_effort","structured_outputs","tool_choice","tools","web_search_options"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-08-28","expiration_date":null,"links":{"details":"/api/v1/models/sakana/fugu-ultra-v2-20260911/endpoints"},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","xhigh","high"],"default_effort":"xhigh"}},{"id":"sakana/fugu-max","canonical_slug":"sakana/fugu-max-20260911","hugging_face_id":null,"name":"Sakana: Fugu Max","created":1789104771,"description":"Fugu Max is the cost-performance model in Sakana AI's Fugu family. Rather than a single monolithic model, Fugu is a learned multi-agent orchestration system: a language model trained to route...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000006","web_search":"0.01","input_cache_read":"0.00000025"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","reasoning","reasoning_effort","structured_outputs","tool_choice","tools","web_search_options"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/sakana/fugu-max-20260911/endpoints"},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","xhigh","high"],"default_effort":"high"}},{"id":"inclusionai/ling-3.0-flash-vl","canonical_slug":"inclusionai/ling-3.0-flash-vl-20260910","hugging_face_id":"inclusionAI/Ling-3.0-flash-VL","name":"inclusionAI: Ling 3.0 Flash VL","created":1789056114,"description":"Ling 3.0 Flash VL builds on Ling 3.0 Flash (124B total / 5.5B active MoE from InclusionAI), further strengthening its language capabilities while adding native visual perception and advanced visual...","context_length":131072,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000006","completion":"0.00000018","input_cache_read":"0.000000012"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/inclusionai/ling-3.0-flash-vl-20260910/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":24.8,"coding_index":57,"agentic_index":30}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"inclusionai/ling-3.0-flash-vl:free","canonical_slug":"inclusionai/ling-3.0-flash-vl-20260910","hugging_face_id":"inclusionAI/Ling-3.0-flash-VL","name":"inclusionAI: Ling 3.0 Flash VL (free)","created":1789056114,"description":"Ling 3.0 Flash VL builds on Ling 3.0 Flash (124B total / 5.5B active MoE from InclusionAI), further strengthening its language capabilities while adding native visual perception and advanced visual...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":262144,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","repetition_penalty","seed","stop","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/inclusionai/ling-3.0-flash-vl-20260910/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":24.8,"coding_index":57,"agentic_index":30}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"deepseek/deepseek-v4.1-flash","canonical_slug":"deepseek/deepseek-v4.1-flash-20260910","hugging_face_id":"deepseek-ai/DeepSeek-V4.1-Flash","name":"DeepSeek: DeepSeek V4.1 Flash","created":1789021285,"description":"DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...","context_length":1048576,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000006","input_cache_read":"0.000000003","overrides":[{"utc_days":["saturday","sunday"],"prompt":"0.00000015","completion":"0.0000006","input_cache_read":"0.000000003"},{"utc_days":["monday","tuesday","wednesday","thursday","friday"],"utc_start":0,"utc_end":100,"prompt":"0.00000015","completion":"0.0000006","input_cache_read":"0.000000003"},{"utc_days":["monday","tuesday","wednesday","thursday","friday"],"utc_start":100,"utc_end":400,"prompt":"0.0000003","completion":"0.0000012","input_cache_read":"0.000000006"},{"utc_days":["monday","tuesday","wednesday","thursday","friday"],"utc_start":400,"utc_end":600,"prompt":"0.00000015","completion":"0.0000006","input_cache_read":"0.000000003"},{"utc_days":["monday","tuesday","wednesday","thursday","friday"],"utc_start":600,"utc_end":1000,"prompt":"0.0000003","completion":"0.0000012","input_cache_read":"0.000000006"},{"utc_days":["monday","tuesday","wednesday","thursday","friday"],"utc_start":1000,"utc_end":0,"prompt":"0.00000015","completion":"0.0000006","input_cache_read":"0.000000003"}]},"top_provider":{"context_length":1048576,"max_completion_tokens":384000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v4.1-flash-20260910/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":39.5,"coding_index":null,"agentic_index":null}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"high"}},{"id":"inception/mercury-2.5","canonical_slug":"inception/mercury-2.5-20260908","hugging_face_id":null,"name":"Inception: Mercury 2.5","created":1788892137,"description":"Mercury 2.5 is the fastest reasoning LLM, and the latest diffusion LLM (dLLM) from Inception. Instead of generating tokens sequentially, Mercury 2.5 produces and refines multiple tokens in parallel, achieving...","context_length":260000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000004","completion":"0.00000015","input_cache_read":"0.000000004"},"top_provider":{"context_length":260000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/inception/mercury-2.5-20260908/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","medium","low","none"],"default_effort":"medium"}},{"id":"nex-agi/nex-n2.5-mini:free","canonical_slug":"nex-agi/nex-n2.5-mini-20260908","hugging_face_id":"nex-agi/Nex-N2.5-mini","name":"Nex AGI: Nex-N2.5-Mini (free)","created":1788890061,"description":"Nex-N2.5 is an agentic model built to turn goals into working, verified outcomes. Its core strength is agentic coding within a visual feedback loop: it can explore codebases, implement multi-file...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","logprobs","max_tokens","reasoning","reasoning_effort","response_format","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.7,"top_p":0.95,"top_k":40},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nex-agi/nex-n2.5-mini-20260908/endpoints"},"reasoning":{"mandatory":false,"supported_efforts":["high","medium","none"],"default_effort":"high"}},{"id":"nex-agi/nex-n2.5-pro:free","canonical_slug":"nex-agi/nex-n2.5-pro-20260907","hugging_face_id":"nex-agi/Nex-N2.5-Pro","name":"Nex AGI: Nex-N2.5-Pro (free)","created":1788890050,"description":"Nex-N2.5 is an agentic model built to turn goals into working, verified outcomes. Its core strength is agentic coding within a visual feedback loop: it can explore codebases, implement multi-file...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","logprobs","max_tokens","reasoning","reasoning_effort","response_format","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.7,"top_p":0.95,"top_k":40},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nex-agi/nex-n2.5-pro-20260907/endpoints"},"reasoning":{"mandatory":false,"supported_efforts":["high","medium","none"],"default_effort":"high"}},{"id":"openai/gpt-6-astra","canonical_slug":"openai/gpt-6-astra-20260903","hugging_face_id":null,"name":"OpenAI: GPT-6 Astra","created":1788552838,"description":"GPT-6 Astra is OpenAI's flagship model for demanding end-to-end work. It is suited for advanced analysis, software engineering, deep research, scientific work, and document creation, with particular strengths in long-horizon...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00001","completion":"0.00005","web_search":"0.01","input_cache_read":"0.000001","input_cache_write":"0.0000125","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00002","completion":"0.000075","input_cache_read":"0.000002","input_cache_write":"0.000025"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-6-astra-20260903/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":52.8,"coding_index":76.9,"agentic_index":51.5}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"medium"}},{"id":"openai/gpt-6-astra:batch","canonical_slug":"openai/gpt-6-astra-20260903","hugging_face_id":null,"name":"OpenAI: GPT-6 Astra (batch)","created":1788552838,"description":"GPT-6 Astra is OpenAI's flagship model for demanding end-to-end work. It is suited for advanced analysis, software engineering, deep research, scientific work, and document creation, with particular strengths in long-horizon...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000025","web_search":"0.01","input_cache_read":"0.0000005","input_cache_write":"0.00000625","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00001","completion":"0.0000375","input_cache_read":"0.000001","input_cache_write":"0.0000125"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-6-astra-20260903/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":52.8,"coding_index":76.9,"agentic_index":51.5}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"medium"}},{"id":"openai/gpt-6-astra-pro","canonical_slug":"openai/gpt-6-astra-pro-20260903","hugging_face_id":null,"name":"OpenAI: GPT-6 Astra Pro","created":1788552835,"description":"GPT-6 Astra Pro is the same underlying model as [GPT-6 Astra](https://openrouter.ai/openai/gpt-6-astra), served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks.\n\nLearn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00001","completion":"0.00005","web_search":"0.01","input_cache_read":"0.000001","input_cache_write":"0.0000125","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00002","completion":"0.000075","input_cache_read":"0.000002","input_cache_write":"0.000025"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-6-astra-pro-20260903/endpoints"},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"medium"}},{"id":"openai/gpt-6-astra-pro:batch","canonical_slug":"openai/gpt-6-astra-pro-20260903","hugging_face_id":null,"name":"OpenAI: GPT-6 Astra Pro (batch)","created":1788552835,"description":"GPT-6 Astra Pro is the same underlying model as [GPT-6 Astra](https://openrouter.ai/openai/gpt-6-astra), served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks.\n\nLearn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000025","web_search":"0.01","input_cache_read":"0.0000005","input_cache_write":"0.00000625","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00001","completion":"0.0000375","input_cache_read":"0.000001","input_cache_write":"0.0000125"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-6-astra-pro-20260903/endpoints"},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"medium"}},{"id":"inclusionai/ling-3.0-flash-sante:free","canonical_slug":"inclusionai/ling-3.0-flash-sante-20260904","hugging_face_id":null,"name":"inclusionAI: Ling 3.0 Flash Sante (free)","created":1788545946,"description":"Ling 3.0 Flash Sante is a health and medicine-focused mixture-of-experts model from InclusionAI, built on Ling 3.0 Flash with 5.1B active parameters out of 124B total. It is designed for...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":262144,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","repetition_penalty","seed","stop","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/inclusionai/ling-3.0-flash-sante-20260904/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"qwen/qwen3.8-max-0902","canonical_slug":"qwen/qwen3.8-max-20260902","hugging_face_id":null,"name":"Qwen: Qwen3.8 Max (0902)","created":1788469704,"description":"Qwen3.8 Max 0902 is an updated snapshot of Qwen3.8 Max from Alibaba's Qwen team. It is a 2.4-trillion-parameter mixture-of-experts model that accepts text, image, and video input and returns text,...","context_length":1000000,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000006","input_cache_read":"0.00000025","input_cache_write":"0.0000025"},"top_provider":{"context_length":1000000,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.8-max-20260902/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":40.3,"coding_index":71.8,"agentic_index":49.6}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["xhigh","high","medium","low","minimal"],"default_effort":"xhigh"}},{"id":"meta/muse-spark-1.3-contributor","canonical_slug":"meta/muse-spark-1.3-contributor-20260902","hugging_face_id":null,"name":"Meta: Muse Spark 1.3 Contributor","created":1788381519,"description":"Muse Spark 1.3 Contributor is the cost-efficient contributor tier of Meta’s multimodal reasoning model for experimentation, learning, and early-stage agentic, multi-agent, and coding workflows. It is designed to track information...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000002","web_search":"0.0025","input_cache_read":"0.000000002"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","repetition_penalty","response_format","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/meta/muse-spark-1.3-contributor-20260902/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["max","xhigh","high","medium","low","minimal"],"default_effort":"medium"}},{"id":"meta/muse-spark-1.3","canonical_slug":"meta/muse-spark-1.3-20260902","hugging_face_id":null,"name":"Meta: Muse Spark 1.3","created":1788378359,"description":"Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It is designed to keep track of information across extended tasks, work through...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.00000425","web_search":"0.0025","input_cache_read":"0.00000015"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","repetition_penalty","response_format","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/meta/muse-spark-1.3-20260902/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"fullstack","elo":1305,"win_rate":56,"rank":5},{"arena":"agents","category":"python-pptxslides","elo":1245,"win_rate":51.5,"rank":10},{"arena":"agents","category":"webapps","elo":1307,"win_rate":56.6,"rank":3},{"arena":"models","category":"3d","elo":1427,"win_rate":64.7,"rank":1},{"arena":"models","category":"codecategories","elo":1373,"win_rate":59.4,"rank":2},{"arena":"models","category":"dataviz","elo":1320,"win_rate":54.2,"rank":8},{"arena":"models","category":"gamedev","elo":1371,"win_rate":58.1,"rank":4},{"arena":"models","category":"uicomponent","elo":1387,"win_rate":60.2,"rank":1},{"arena":"models","category":"website","elo":1364,"win_rate":59.5,"rank":1}]},"reasoning":{"mandatory":true,"supported_efforts":["max","xhigh","high","medium","low","minimal"],"default_effort":"medium"}},{"id":"google/gemini-3.8-flash","canonical_slug":"google/gemini-3.8-flash-20260902","hugging_face_id":null,"name":"Google: Gemini 3.8 Flash","created":1788362056,"description":"Gemini 3.8 Flash is Google's most intelligent Flash model with significant gains from 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning.","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000075","completion":"0.00000375","image":"0.00000075","audio":"0.00000075","input_audio_cache":"0.000000075","web_search":"0.014","internal_reasoning":"0.00000375","input_cache_read":"0.000000075","input_cache_write":"0.0000000416666666666667"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.8-flash-20260902/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"fullstack","elo":1245,"win_rate":48.4,"rank":12},{"arena":"agents","category":"mobileapps","elo":1281,"win_rate":58.8,"rank":3},{"arena":"agents","category":"python-pptxslides","elo":1176,"win_rate":39.5,"rank":19},{"arena":"agents","category":"webapps","elo":1246,"win_rate":45.8,"rank":9},{"arena":"models","category":"3d","elo":1323,"win_rate":52.2,"rank":12},{"arena":"models","category":"codecategories","elo":1323,"win_rate":53.3,"rank":9},{"arena":"models","category":"dataviz","elo":1260,"win_rate":49.3,"rank":27},{"arena":"models","category":"gamedev","elo":1343,"win_rate":55.2,"rank":8},{"arena":"models","category":"uicomponent","elo":1334,"win_rate":54.1,"rank":8},{"arena":"models","category":"website","elo":1315,"win_rate":52.9,"rank":8}],"artificial_analysis":{"intelligence_index":41.2,"coding_index":76.3,"agentic_index":41.1}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"google/gemini-3.8-flash:batch","canonical_slug":"google/gemini-3.8-flash-20260902","hugging_face_id":null,"name":"Google: Gemini 3.8 Flash (batch)","created":1788362056,"description":"Gemini 3.8 Flash is Google's most intelligent Flash model with significant gains from 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning.","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.000000375","completion":"0.000001875","image":"0.000000375","audio":"0.000000375","input_audio_cache":"0.0000000375","web_search":"0.014","internal_reasoning":"0.000001875","input_cache_read":"0.0000000375","input_cache_write":"0.0000000416666666666667"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.8-flash-20260902/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"fullstack","elo":1245,"win_rate":48.4,"rank":12},{"arena":"agents","category":"mobileapps","elo":1281,"win_rate":58.8,"rank":3},{"arena":"agents","category":"python-pptxslides","elo":1176,"win_rate":39.5,"rank":19},{"arena":"agents","category":"webapps","elo":1246,"win_rate":45.8,"rank":9},{"arena":"models","category":"3d","elo":1323,"win_rate":52.2,"rank":12},{"arena":"models","category":"codecategories","elo":1323,"win_rate":53.3,"rank":9},{"arena":"models","category":"dataviz","elo":1260,"win_rate":49.3,"rank":27},{"arena":"models","category":"gamedev","elo":1343,"win_rate":55.2,"rank":8},{"arena":"models","category":"uicomponent","elo":1334,"win_rate":54.1,"rank":8},{"arena":"models","category":"website","elo":1315,"win_rate":52.9,"rank":8}],"artificial_analysis":{"intelligence_index":41.2,"coding_index":76.3,"agentic_index":41.1}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"anthropic/claude-fable-5.1","canonical_slug":"anthropic/claude-fable-5.1-20260831","hugging_face_id":null,"name":"Anthropic: Claude Fable 5.1","created":1788285838,"description":"Claude Fable 5.1 improves on Claude Fable 5 across the board, with the biggest gains in agentic coding, long-running agentic workflows, and knowledge work: long code refactors, front-end and visual...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.00001","completion":"0.00005","web_search":"0.01","input_cache_read":"0.00000025","input_cache_write":"0.0000125","input_cache_write_1h":"0.00002"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-fable-5.1-20260831/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"fullstack","elo":1338,"win_rate":59.6,"rank":1},{"arena":"agents","category":"python-pptxslides","elo":1328,"win_rate":62.6,"rank":1},{"arena":"models","category":"3d","elo":1423,"win_rate":68.5,"rank":3},{"arena":"models","category":"asciiart","elo":1324,"win_rate":64.9,"rank":3},{"arena":"models","category":"codecategories","elo":1345,"win_rate":59.1,"rank":3},{"arena":"models","category":"dataviz","elo":1362,"win_rate":62.9,"rank":3},{"arena":"models","category":"gamedev","elo":1415,"win_rate":67.1,"rank":1},{"arena":"models","category":"svg","elo":1351,"win_rate":64.3,"rank":1},{"arena":"models","category":"uicomponent","elo":1335,"win_rate":58.1,"rank":7},{"arena":"models","category":"website","elo":1322,"win_rate":56.3,"rank":3}],"artificial_analysis":{"intelligence_index":53.4,"coding_index":81.6,"agentic_index":58}},"reasoning":{"mandatory":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"anthropic/claude-fable-5.1:batch","canonical_slug":"anthropic/claude-fable-5.1-20260831","hugging_face_id":null,"name":"Anthropic: Claude Fable 5.1 (batch)","created":1788285838,"description":"Claude Fable 5.1 improves on Claude Fable 5 across the board, with the biggest gains in agentic coding, long-running agentic workflows, and knowledge work: long code refactors, front-end and visual...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000025","web_search":"0.01","input_cache_read":"0.000000125","input_cache_write":"0.00000625","input_cache_write_1h":"0.00001"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-fable-5.1-20260831/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"fullstack","elo":1338,"win_rate":59.6,"rank":1},{"arena":"agents","category":"python-pptxslides","elo":1328,"win_rate":62.6,"rank":1},{"arena":"models","category":"3d","elo":1423,"win_rate":68.5,"rank":3},{"arena":"models","category":"asciiart","elo":1324,"win_rate":64.9,"rank":3},{"arena":"models","category":"codecategories","elo":1345,"win_rate":59.1,"rank":3},{"arena":"models","category":"dataviz","elo":1362,"win_rate":62.9,"rank":3},{"arena":"models","category":"gamedev","elo":1415,"win_rate":67.1,"rank":1},{"arena":"models","category":"svg","elo":1351,"win_rate":64.3,"rank":1},{"arena":"models","category":"uicomponent","elo":1335,"win_rate":58.1,"rank":7},{"arena":"models","category":"website","elo":1322,"win_rate":56.3,"rank":3}],"artificial_analysis":{"intelligence_index":53.4,"coding_index":81.6,"agentic_index":58}},"reasoning":{"mandatory":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"ibm-granite/granite-4.2-8b","canonical_slug":"ibm-granite/granite-4.2-8b-20260831","hugging_face_id":"ibm-granite/granite-4.2-8b","name":"IBM: Granite 4.2 8B","created":1788206780,"description":"Granite 4.2 8B is a dense reasoning model from IBM. It is suited for mathematics, code generation, multilingual dialogue, and agentic workflows that need multi-step reasoning. It supports full, low-effort,...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000006","completion":"0.00000025","input_cache_read":"0.000000015"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/ibm-granite/granite-4.2-8b-20260831/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":11.8,"coding_index":22.4,"agentic_index":3.7}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","low","none"],"default_effort":"high"}},{"id":"tencent/hy4-preview","canonical_slug":"tencent/hy4-preview-20260827","hugging_face_id":"tencent/Hy4-preview","name":"Tencent: Hy4 preview","created":1787897375,"description":"Tencent: Hy4 preview is a mixture-of-experts model from Tencent, with 49B active parameters out of 770B total. It is designed for coding agents, complex tool-use workflows, and productivity tasks that...","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000834","completion":"0.000002501","input_cache_read":"0.000000042"},"top_provider":{"context_length":1048576,"max_completion_tokens":64000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/tencent/hy4-preview-20260827/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","low","none"],"default_effort":"high"}},{"id":"inclusionai/ling-3.0-flash-fin","canonical_slug":"inclusionai/ling-3.0-flash-fin-20260827","hugging_face_id":null,"name":"inclusionAI: Ling 3.0 Flash Fin","created":1787846290,"description":"Ling 3.0 Flash Fin is a finance-focused mixture-of-experts model from InclusionAI, built on Ling 3.0 Flash with 5.1B active parameters out of 124B total. It is designed for real-world investment...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000006","completion":"0.00000018","input_cache_read":"0.000000012"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/inclusionai/ling-3.0-flash-fin-20260827/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"inclusionai/ling-3.0-flash-fin:free","canonical_slug":"inclusionai/ling-3.0-flash-fin-20260827","hugging_face_id":null,"name":"inclusionAI: Ling 3.0 Flash Fin (free)","created":1787846290,"description":"Ling 3.0 Flash Fin is a finance-focused mixture-of-experts model from InclusionAI, built on Ling 3.0 Flash with 5.1B active parameters out of 124B total. It is designed for real-world investment...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":262144,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","repetition_penalty","seed","stop","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/inclusionai/ling-3.0-flash-fin-20260827/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"~z-ai/glm-flash-latest","canonical_slug":"~z-ai/glm-flash-latest","alias_target":{"name":"Z.ai: GLM 5.3 Flash","slug":"z-ai/glm-5.3-flash"},"hugging_face_id":null,"name":"Z.ai: GLM Flash Latest","created":1787817633,"description":"This model always redirects to the latest model in the GLM Flash family.","context_length":1310720,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.000000075","completion":"0.00000025","input_cache_read":"0.000000015"},"top_provider":{"context_length":1048576,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~z-ai/glm-flash-latest/endpoints"},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"max"}},{"id":"qwen/qwen3.8-flash","canonical_slug":"qwen/qwen3.8-flash-20260826","hugging_face_id":"Qwen/Qwen3.8-Flash-Next","name":"Qwen: Qwen3.8 Flash","created":1787773060,"description":"Qwen3.8 Flash is a multimodal reasoning model from Alibaba. It is suited for coding assistance, agentic workflows, visual understanding, document and codebase analysis, desktop interaction, chart analysis, and long-video analysis.","context_length":1000000,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.00000047","input_cache_read":"0.000000016","input_cache_write":"0.0000002"},"top_provider":{"context_length":1000000,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.8-flash-20260826/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supports_max_tokens":true}},{"id":"z-ai/glm-5.3-flash","canonical_slug":"z-ai/glm-5.3-flash-20260826","hugging_face_id":"zai-org/GLM-5.3-Flash","name":"Z.ai: GLM 5.3 Flash","created":1787752741,"description":"GLM-5.3-Flash is a native multimodal model from Z.ai. It is suited for efficient coding and long-horizon agent tasks. Its hybrid sparse and linear attention architecture maintains accurate long-context behavior while...","context_length":1310720,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000005","input_cache_read":"0.00000003"},"top_provider":{"context_length":1048576,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":"2098-12-31","links":{"details":"/api/v1/models/z-ai/glm-5.3-flash-20260826/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1356,"win_rate":60.3,"rank":7},{"arena":"models","category":"asciiart","elo":1293,"win_rate":56.7,"rank":7},{"arena":"models","category":"codecategories","elo":1298,"win_rate":49.8,"rank":15},{"arena":"models","category":"dataviz","elo":1275,"win_rate":50.7,"rank":21},{"arena":"models","category":"gamedev","elo":1307,"win_rate":47.3,"rank":14},{"arena":"models","category":"svg","elo":1313,"win_rate":57.5,"rank":7},{"arena":"models","category":"uicomponent","elo":1344,"win_rate":57.1,"rank":5},{"arena":"models","category":"website","elo":1285,"win_rate":48.5,"rank":20}],"artificial_analysis":{"intelligence_index":41.9,"coding_index":71.5,"agentic_index":51.2}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"max"}},{"id":"z-ai/glm-5.3-flash:batch","canonical_slug":"z-ai/glm-5.3-flash-20260826","hugging_face_id":"zai-org/GLM-5.3-Flash","name":"Z.ai: GLM 5.3 Flash (batch)","created":1787752741,"description":"GLM-5.3-Flash is a native multimodal model from Z.ai. It is suited for efficient coding and long-horizon agent tasks. Its hybrid sparse and linear attention architecture maintains accurate long-context behavior while...","context_length":1048576,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000075","completion":"0.00000025","input_cache_read":"0.000000015"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-5.3-flash-20260826/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1356,"win_rate":60.3,"rank":7},{"arena":"models","category":"asciiart","elo":1293,"win_rate":56.7,"rank":7},{"arena":"models","category":"codecategories","elo":1298,"win_rate":49.8,"rank":15},{"arena":"models","category":"dataviz","elo":1275,"win_rate":50.7,"rank":21},{"arena":"models","category":"gamedev","elo":1307,"win_rate":47.3,"rank":14},{"arena":"models","category":"svg","elo":1313,"win_rate":57.5,"rank":7},{"arena":"models","category":"uicomponent","elo":1344,"win_rate":57.1,"rank":5},{"arena":"models","category":"website","elo":1285,"win_rate":48.5,"rank":20}],"artificial_analysis":{"intelligence_index":41.9,"coding_index":71.5,"agentic_index":51.2}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"max"}},{"id":"meta/muse-spark-1.2-contributor","canonical_slug":"meta/muse-spark-1.2-contributor-20260805","hugging_face_id":null,"name":"Meta: Muse Spark 1.2 Contributor","created":1787336476,"description":"Muse Spark 1.2 contributor tier is a reasoning model from Meta designed for developers who want to start building at an even lower cost. It’s meaningfully cheaper than Muse Spark...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000002","web_search":"0.0025","input_cache_read":"0.000000002"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","repetition_penalty","response_format","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/meta/muse-spark-1.2-contributor-20260805/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium","low","minimal"],"default_effort":"medium"}},{"id":"deepseek/deepseek-v4-flash-vision-exp","canonical_slug":"deepseek/deepseek-v4-flash-vision-exp-20260821","hugging_face_id":"deepseek-ai/DeepSeek-V4-Flash-Vision-Exp","name":"DeepSeek: DeepSeek V4 Flash Vision Exp","created":1787311563,"description":"DeepSeek V4 Flash Vision Exp is an experimental vision-enabled version of [DeepSeek V4 Flash 0731](https://openrouter.ai/deepseek/deepseek-v4-flash-0731) from DeepSeek, adding image understanding while matching the base model on text capabilities including agents,...","context_length":1048576,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.00000022","completion":"0.00000066","input_cache_read":"0.000000007"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v4-flash-vision-exp-20260821/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"high"}},{"id":"deepseek/deepseek-v4-flash-vision-exp:batch","canonical_slug":"deepseek/deepseek-v4-flash-vision-exp-20260821","hugging_face_id":"deepseek-ai/DeepSeek-V4-Flash-Vision-Exp","name":"DeepSeek: DeepSeek V4 Flash Vision Exp (batch)","created":1787311563,"description":"DeepSeek V4 Flash Vision Exp is an experimental vision-enabled version of [DeepSeek V4 Flash 0731](https://openrouter.ai/deepseek/deepseek-v4-flash-0731) from DeepSeek, adding image understanding while matching the base model on text capabilities including agents,...","context_length":1048576,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.00000011","completion":"0.00000033","input_cache_read":"0.0000000035"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v4-flash-vision-exp-20260821/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"high"}},{"id":"tencent/hy-mt2-1.8b","canonical_slug":"tencent/hy-mt2-1.8b-20260521","hugging_face_id":"tencent/Hy-MT2-1.8B","name":"Tencent: Hy-MT2-1.8B","created":1787231581,"description":"Hy-MT2-1.8B is a compact 1.8B-parameter translation model from Tencent. It supports 33 language pairs and five Chinese dialect and minority-language pairs, with workflows for structured, delimiter-based, contextual, glossary-based, and style-guided...","context_length":8192,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000044","completion":"0.000000177"},"top_provider":{"context_length":8192,"max_completion_tokens":4096,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_completion_tokens","max_tokens","stop","temperature"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/tencent/hy-mt2-1.8b-20260521/endpoints"}},{"id":"tencent/hy-mt2-30b-a3b","canonical_slug":"tencent/hy-mt2-30b-a3b-20260521","hugging_face_id":"tencent/Hy-MT2-30B-A3B","name":"Tencent: Hy-MT2-30B-A3B","created":1787231561,"description":"Hy-MT2-30B-A3B is Tencent's flagship translation model in the Hy-MT2 family. It supports 33 language pairs and five Chinese dialect and minority-language pairs, with workflows for structured, delimiter-based, contextual, glossary-based, and...","context_length":8192,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000074","completion":"0.000000295"},"top_provider":{"context_length":8192,"max_completion_tokens":4096,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_completion_tokens","max_tokens","response_format","stop","structured_outputs","temperature"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/tencent/hy-mt2-30b-a3b-20260521/endpoints"}},{"id":"~z-ai/glm-latest","canonical_slug":"~z-ai/glm-latest","alias_target":{"name":"Z.ai: GLM 5.3","slug":"z-ai/glm-5.3"},"hugging_face_id":null,"name":"Z.ai: GLM Latest","created":1787151053,"description":"This model always redirects to the latest GLM model from Z.ai.","context_length":1310720,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.0000009","completion":"0.000003069","input_cache_read":"0.00000018"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","parallel_tool_calls","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~z-ai/glm-latest/endpoints"},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"max"}},{"id":"tencent/hy-mt2-7b","canonical_slug":"tencent/hy-mt2-7b-20260521","hugging_face_id":"tencent/Hy-MT2-7B","name":"Tencent: Hy-MT2-7B","created":1787148797,"description":"Hy-MT2-7B is a 7B-parameter translation model from Tencent. It supports 33 language pairs and five Chinese dialect and minority-language pairs, with workflows for structured, delimiter-based, contextual, glossary-based, and style-guided translation.","context_length":8192,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000074","completion":"0.000000295"},"top_provider":{"context_length":8192,"max_completion_tokens":4096,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_completion_tokens","max_tokens","response_format","stop","structured_outputs","temperature"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/tencent/hy-mt2-7b-20260521/endpoints"}},{"id":"z-ai/glm-5.3","canonical_slug":"z-ai/glm-5.3-20260816","hugging_face_id":"zai-org/GLM-5.3","name":"Z.ai: GLM 5.3","created":1787086655,"description":"GLM-5.3 is a large-scale reasoning model from Z.ai, built for complex software engineering and long-horizon agent tasks. It supports text input and output with a 1M-token context window, and improves...","context_length":1310720,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000014","completion":"0.0000044","input_cache_read":"0.00000026"},"top_provider":{"context_length":1048575,"max_completion_tokens":943717,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","parallel_tool_calls","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-5.3-20260816/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1393,"win_rate":67.8,"rank":4},{"arena":"models","category":"codecategories","elo":1331,"win_rate":58.1,"rank":5},{"arena":"models","category":"dataviz","elo":1262,"win_rate":51,"rank":25},{"arena":"models","category":"gamedev","elo":1376,"win_rate":64.2,"rank":3},{"arena":"models","category":"svg","elo":1325,"win_rate":60.1,"rank":5},{"arena":"models","category":"uicomponent","elo":1343,"win_rate":59.5,"rank":6},{"arena":"models","category":"website","elo":1318,"win_rate":56.3,"rank":6},{"arena":"agents","category":"htmlslides","elo":1189,"win_rate":39.1,"rank":9},{"arena":"agents","category":"mobileapps","elo":1229,"win_rate":54.4,"rank":13},{"arena":"agents","category":"python-pptxslides","elo":1258,"win_rate":51.6,"rank":7}],"artificial_analysis":{"intelligence_index":44.9,"coding_index":74.8,"agentic_index":53.4}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"max"}},{"id":"z-ai/glm-5.3:batch","canonical_slug":"z-ai/glm-5.3-20260816","hugging_face_id":"zai-org/GLM-5.3","name":"Z.ai: GLM 5.3 (batch)","created":1787086655,"description":"GLM-5.3 is a large-scale reasoning model from Z.ai, built for complex software engineering and long-horizon agent tasks. It supports text input and output with a 1M-token context window, and improves...","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000007","completion":"0.0000022","input_cache_read":"0.00000013"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-5.3-20260816/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1393,"win_rate":67.8,"rank":4},{"arena":"models","category":"codecategories","elo":1331,"win_rate":58.1,"rank":5},{"arena":"models","category":"dataviz","elo":1262,"win_rate":51,"rank":25},{"arena":"models","category":"gamedev","elo":1376,"win_rate":64.2,"rank":3},{"arena":"models","category":"svg","elo":1325,"win_rate":60.1,"rank":5},{"arena":"models","category":"uicomponent","elo":1343,"win_rate":59.5,"rank":6},{"arena":"models","category":"website","elo":1318,"win_rate":56.3,"rank":6},{"arena":"agents","category":"htmlslides","elo":1189,"win_rate":39.1,"rank":9},{"arena":"agents","category":"mobileapps","elo":1229,"win_rate":54.4,"rank":13},{"arena":"agents","category":"python-pptxslides","elo":1258,"win_rate":51.6,"rank":7}],"artificial_analysis":{"intelligence_index":44.9,"coding_index":74.8,"agentic_index":53.4}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"max"}},{"id":"qwen/qwen3.8-27b","canonical_slug":"qwen/qwen3.8-27b-20260814","hugging_face_id":"Qwen/Qwen3.8-27B","name":"Qwen: Qwen3.8 27B","created":1786722910,"description":"Qwen3.8 27B is an open-weight dense vision-language model from Qwen. It is suited for coding, professional workflows, research, multimodal interaction, and long-running agent tasks, with flexible thinking that can be...","context_length":1000000,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.000000214","completion":"0.00000255","input_cache_read":"0.00000015"},"top_provider":{"context_length":262144,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_a","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":20},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.8-27b-20260814/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":33.9,"coding_index":68.1,"agentic_index":46.5}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["xhigh","medium","low"],"default_effort":"xhigh"}},{"id":"dots-studio/dots-3-note-preview:free","canonical_slug":"dots-studio/dots-3-note-preview-20260813","hugging_face_id":null,"name":"Dots Studio: Dots3-Note Preview (free)","created":1786680361,"description":"Dots3-Note Preview is an open-weight mixture-of-experts model from Dots Studio, with 16B active parameters out of 280B total. It is the lightest model in the Dots 3 family and is...","context_length":512000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":512000,"max_completion_tokens":460800,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":"2026-09-30","links":{"details":"/api/v1/models/dots-studio/dots-3-note-preview-20260813/endpoints"},"reasoning":{"mandatory":false}},{"id":"google/gemini-3.7-flash","canonical_slug":"google/gemini-3.7-flash-20260813","hugging_face_id":null,"name":"Google: Gemini 3.7 Flash","created":1786640581,"description":"Gemini 3.7 Flash is a multimodal model from Google for fast agentic workflows, coding, and complex multi-step reasoning. It is designed for tasks that require responsive performance and reliable multi-step...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000075","completion":"0.00000375","image":"0.00000075","audio":"0.00000075","input_audio_cache":"0.000000075","web_search":"0.014","internal_reasoning":"0.00000375","input_cache_read":"0.000000075","input_cache_write":"0.0000000416666666666667"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.7-flash-20260813/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1235,"win_rate":52.1,"rank":5},{"arena":"agents","category":"androidnative","elo":1256,"win_rate":53.6,"rank":5},{"arena":"agents","category":"fullstack","elo":1225,"win_rate":46.2,"rank":14},{"arena":"agents","category":"mobileapps","elo":1263,"win_rate":54.3,"rank":5},{"arena":"agents","category":"webapps","elo":1234,"win_rate":47.9,"rank":12},{"arena":"models","category":"3d","elo":1339,"win_rate":57.3,"rank":9},{"arena":"models","category":"asciiart","elo":1264,"win_rate":51.1,"rank":15},{"arena":"models","category":"codecategories","elo":1323,"win_rate":56.5,"rank":8},{"arena":"models","category":"dataviz","elo":1337,"win_rate":58.8,"rank":6},{"arena":"models","category":"gamedev","elo":1349,"win_rate":56.4,"rank":7},{"arena":"models","category":"uicomponent","elo":1314,"win_rate":52.5,"rank":13},{"arena":"models","category":"website","elo":1315,"win_rate":56.5,"rank":7}],"artificial_analysis":{"intelligence_index":39.4,"coding_index":76.1,"agentic_index":36.4}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"google/gemini-3.7-flash:batch","canonical_slug":"google/gemini-3.7-flash-20260813","hugging_face_id":null,"name":"Google: Gemini 3.7 Flash (batch)","created":1786640581,"description":"Gemini 3.7 Flash is a multimodal model from Google for fast agentic workflows, coding, and complex multi-step reasoning. It is designed for tasks that require responsive performance and reliable multi-step...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.000000375","completion":"0.000001875","image":"0.000000375","audio":"0.000000375","input_audio_cache":"0.0000000375","web_search":"0.014","internal_reasoning":"0.000001875","input_cache_read":"0.0000000375","input_cache_write":"0.0000000416666666666667"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.7-flash-20260813/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1235,"win_rate":52.1,"rank":5},{"arena":"agents","category":"androidnative","elo":1256,"win_rate":53.6,"rank":5},{"arena":"agents","category":"fullstack","elo":1225,"win_rate":46.2,"rank":14},{"arena":"agents","category":"mobileapps","elo":1263,"win_rate":54.3,"rank":5},{"arena":"agents","category":"webapps","elo":1234,"win_rate":47.9,"rank":12},{"arena":"models","category":"3d","elo":1339,"win_rate":57.3,"rank":9},{"arena":"models","category":"asciiart","elo":1264,"win_rate":51.1,"rank":15},{"arena":"models","category":"codecategories","elo":1323,"win_rate":56.5,"rank":8},{"arena":"models","category":"dataviz","elo":1337,"win_rate":58.8,"rank":6},{"arena":"models","category":"gamedev","elo":1349,"win_rate":56.4,"rank":7},{"arena":"models","category":"uicomponent","elo":1314,"win_rate":52.5,"rank":13},{"arena":"models","category":"website","elo":1315,"win_rate":56.5,"rank":7}],"artificial_analysis":{"intelligence_index":39.4,"coding_index":76.1,"agentic_index":36.4}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"bytedance-seed/seed-2-1-turbo","canonical_slug":"bytedance-seed/seed-2-1-turbo-20260810","hugging_face_id":null,"name":"ByteDance Seed: Seed 2.1 Turbo","created":1786552176,"description":"Seed 2.1 Turbo is a multimodal model from ByteDance Seed for coding and long-horizon agent workflows. It is suited for end-to-end software delivery, multi-step task execution, and understanding visual and...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000005","completion":"0.0000025"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","reasoning","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/bytedance-seed/seed-2-1-turbo-20260810/endpoints"},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3.8-2.4t-a95b","canonical_slug":"qwen/qwen3.8-2.4t-a95b-20260812","hugging_face_id":"Qwen/Qwen3.8-2.4T-A95B","name":"Qwen: Qwen3.8 2.4T A95B","created":1786551702,"description":"Qwen3.8 2.4T A95B is an open-weight sparse mixture-of-experts model from Qwen and the open-weight variant of [Qwen3.8 Max](/qwen/qwen3.8-max), with 95 billion active parameters out of 2.4 trillion total. It is...","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000006","input_cache_read":"0.00000025"},"top_provider":{"context_length":1000000,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":20},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.8-2.4t-a95b-20260812/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":40,"coding_index":71.9,"agentic_index":50.4}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["xhigh","medium","low"],"default_effort":"xhigh"}},{"id":"qwen/qwen3.8-2.4t-a95b:batch","canonical_slug":"qwen/qwen3.8-2.4t-a95b-20260812","hugging_face_id":"Qwen/Qwen3.8-2.4T-A95B","name":"Qwen: Qwen3.8 2.4T A95B (batch)","created":1786551702,"description":"Qwen3.8 2.4T A95B is an open-weight sparse mixture-of-experts model from Qwen and the open-weight variant of [Qwen3.8 Max](/qwen/qwen3.8-max), with 95 billion active parameters out of 2.4 trillion total. It is...","context_length":1010000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000006","input_cache_read":"0.00000025"},"top_provider":{"context_length":1010000,"max_completion_tokens":909000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":20},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.8-2.4t-a95b-20260812/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":40,"coding_index":71.9,"agentic_index":50.4}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["xhigh","medium","low"],"default_effort":"xhigh"}},{"id":"bytedance-seed/seed-2.0-code","canonical_slug":"bytedance-seed/seed-2.0-code-20260730","hugging_face_id":null,"name":"ByteDance Seed: Seed-2.0-Code","created":1786550701,"description":"Seed 2.0 Code is a model from ByteDance Seed optimized for agentic coding. It is suited for frontend development, multilingual programming tasks, and coding-agent workflows in tools such as Claude...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000005","completion":"0.000003","overrides":[{"min_prompt_tokens":128000,"prompt":"0.000001","completion":"0.000006"}]},"top_provider":{"context_length":262144,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/bytedance-seed/seed-2.0-code-20260730/endpoints"},"reasoning":{"mandatory":false,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"deepseek/deepseek-v4-pro-0813","canonical_slug":"deepseek/deepseek-v4-pro-20260813","hugging_face_id":"deepseek-ai/DeepSeek-V4-Pro-0813","name":"DeepSeek: DeepSeek V4 Pro 0813","created":1786549364,"description":"DeepSeek V4 Pro 0813 is a large-scale mixture-of-experts model from DeepSeek. This is the GA release of DeepSeek V4 Pro.","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.00000057948","completion":"0.00000173844","input_cache_read":"0.000000018438"},"top_provider":{"context_length":1048576,"max_completion_tokens":393216,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":1},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v4-pro-20260813/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":36.3,"coding_index":68.8,"agentic_index":42.3}},"reasoning":{"mandatory":false,"supported_efforts":["max","high","low"],"default_effort":"high"}},{"id":"deepseek/deepseek-v4-pro-0813:batch","canonical_slug":"deepseek/deepseek-v4-pro-20260813","hugging_face_id":"deepseek-ai/DeepSeek-V4-Pro-0813","name":"DeepSeek: DeepSeek V4 Pro 0813 (batch)","created":1786549364,"description":"DeepSeek V4 Pro 0813 is a large-scale mixture-of-experts model from DeepSeek. This is the GA release of DeepSeek V4 Pro.","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.00000066","completion":"0.00000198","input_cache_read":"0.000000022"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":1},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v4-pro-20260813/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":36.3,"coding_index":68.8,"agentic_index":42.3}},"reasoning":{"mandatory":false,"supported_efforts":["max","high","low"],"default_effort":"high"}},{"id":"x-ai/grok-4.6","canonical_slug":"x-ai/grok-4.6-20260810","hugging_face_id":null,"name":"SpaceXAI: Grok 4.6","created":1786548957,"description":"Grok 4.6 is SpaceXAI's smartest model with frontier performance on coding, knowledge work, and STEM.","context_length":500000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Grok","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000006","web_search":"0.005","input_cache_read":"0.0000005","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000004","completion":"0.000012","input_cache_read":"0.000001"}]},"top_provider":{"context_length":500000,"max_completion_tokens":450000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","logprobs","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/x-ai/grok-4.6-20260810/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1210,"win_rate":47.2,"rank":9},{"arena":"agents","category":"androidnative","elo":1287,"win_rate":59,"rank":2},{"arena":"agents","category":"fullstack","elo":1274,"win_rate":55.6,"rank":7},{"arena":"agents","category":"htmlslides","elo":1253,"win_rate":60.3,"rank":3},{"arena":"agents","category":"mobileapps","elo":1263,"win_rate":57.5,"rank":6},{"arena":"agents","category":"python-pptxslides","elo":1241,"win_rate":52.6,"rank":11},{"arena":"agents","category":"webapps","elo":1260,"win_rate":56.8,"rank":7},{"arena":"models","category":"3d","elo":1306,"win_rate":52.8,"rank":14},{"arena":"models","category":"asciiart","elo":1291,"win_rate":58.3,"rank":9},{"arena":"models","category":"codecategories","elo":1309,"win_rate":53,"rank":11},{"arena":"models","category":"dataviz","elo":1304,"win_rate":51.9,"rank":11},{"arena":"models","category":"gamedev","elo":1322,"win_rate":54.7,"rank":11},{"arena":"models","category":"svg","elo":1264,"win_rate":51.4,"rank":11},{"arena":"models","category":"uicomponent","elo":1315,"win_rate":55.3,"rank":12},{"arena":"models","category":"website","elo":1305,"win_rate":52.9,"rank":11}],"artificial_analysis":{"intelligence_index":44.4,"coding_index":76.8,"agentic_index":53.4}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["xhigh","high","medium","low"],"default_effort":"high"}},{"id":"liquid/lfm-2.5-2.6b:free","canonical_slug":"liquid/lfm-2.5-2.6b-20260811","hugging_face_id":"LiquidAI/LFM2.5-2.6B","name":"LiquidAI: LFM2.5-2.6B (free)","created":1786470519,"description":"LFM2.5-2.6B is a compact reasoning model from Liquid AI. It is suited for agent workflows, data extraction, RAG, and long-context processing. Liquid advises against using it for agentic coding or...","context_length":65536,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":65536,"max_completion_tokens":8192,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_completion_tokens","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs"],"default_parameters":{"temperature":0.1,"top_k":50,"repetition_penalty":1.1},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/liquid/lfm-2.5-2.6b-20260811/endpoints"},"reasoning":{"mandatory":true}},{"id":"nvidia/nemotron-3.5-lightning","canonical_slug":"nvidia/nemotron-3.5-lightning-20260807","hugging_face_id":"nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16","name":"NVIDIA: Nemotron 3.5 Lightning","created":1786452751,"description":"NVIDIA Nemotron 3.5 Lightning is an open mixture-of-experts model from NVIDIA, with 3B active parameters out of 30B total. It is suited for high-throughput agentic workloads and specialized tasks that...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000008","completion":"0.0000002","input_cache_read":"0.00000004"},"top_provider":{"context_length":262144,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nvidia/nemotron-3.5-lightning-20260807/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":13.6,"coding_index":26.8,"agentic_index":6.1}},"reasoning":{"mandatory":false}},{"id":"nvidia/nemotron-3.5-lightning:free","canonical_slug":"nvidia/nemotron-3.5-lightning-20260807","hugging_face_id":"nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16","name":"NVIDIA: Nemotron 3.5 Lightning (free)","created":1786452751,"description":"NVIDIA Nemotron 3.5 Lightning is an open mixture-of-experts model from NVIDIA, with 3B active parameters out of 30B total. It is suited for high-throughput agentic workloads and specialized tasks that...","context_length":1000000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":1000000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","seed","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nvidia/nemotron-3.5-lightning-20260807/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":13.6,"coding_index":26.8,"agentic_index":6.1}},"reasoning":{"mandatory":false}},{"id":"sakana/sakana-namazu","canonical_slug":"sakana/namazu-20260811","hugging_face_id":null,"name":"Sakana: Sakana Namazu","created":1786410129,"description":"Sakana Namazu is a Japanese-specialized reasoning model from Sakana AI, based on Kimi K2.6 with additional training for Japanese language and business contexts. It is suited for Japanese instruction following,...","context_length":262144,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000095","completion":"0.000004","web_search":"0.007","input_cache_read":"0.00000015"},"top_provider":{"context_length":262144,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","reasoning","reasoning_effort","structured_outputs","tool_choice","tools","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/sakana/namazu-20260811/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","none"],"default_effort":"high"}},{"id":"upstage/solar-pro4","canonical_slug":"upstage/solar-pro4-20260810","hugging_face_id":null,"name":"Upstage: Solar Pro 4","created":1786371636,"description":"Solar Pro 4 is Upstage's cost-efficient large language model, featuring a 524K context window. It is built for long-horizon tasks and agentic workflows, with strong capabilities in office productivity, document-intensive...","context_length":524288,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000009","completion":"0.00000036","input_cache_read":"0.000000018"},"top_provider":{"context_length":524288,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","parallel_tool_calls","presence_penalty","reasoning","response_format","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/upstage/solar-pro4-20260810/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"webapps","elo":1118,"win_rate":30.2,"rank":32},{"arena":"models","category":"3d","elo":1207,"win_rate":46.4,"rank":46},{"arena":"models","category":"codecategories","elo":1197,"win_rate":38.9,"rank":54},{"arena":"models","category":"dataviz","elo":1196,"win_rate":41.5,"rank":56},{"arena":"models","category":"gamedev","elo":1194,"win_rate":39.9,"rank":54},{"arena":"models","category":"uicomponent","elo":1166,"win_rate":35.5,"rank":66},{"arena":"models","category":"website","elo":1188,"win_rate":37.4,"rank":65}],"artificial_analysis":{"intelligence_index":null,"coding_index":52.7,"agentic_index":null}},"reasoning":{"mandatory":false}},{"id":"meta/muse-glimmer-30b","canonical_slug":"meta/muse-glimmer-30b-20260810","hugging_face_id":"meta-models/Muse-Glimmer-30B","name":"Meta: Muse Glimmer 30B","created":1786302394,"description":"Muse Glimmer 30B is a dense, open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware. It is suited for long-horizon...","context_length":131072,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000035","completion":"0.0000015","input_cache_read":"0.00000004"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":64},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/meta/muse-glimmer-30b-20260810/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium","low"],"default_effort":"medium"}},{"id":"meta/muse-glimmer-30b:batch","canonical_slug":"meta/muse-glimmer-30b-20260810","hugging_face_id":"meta-models/Muse-Glimmer-30B","name":"Meta: Muse Glimmer 30B (batch)","created":1786302394,"description":"Muse Glimmer 30B is a dense, open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware. It is suited for long-horizon...","context_length":131072,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000175","completion":"0.00000075","input_cache_read":"0.00000002"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":64},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/meta/muse-glimmer-30b-20260810/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium","low"],"default_effort":"medium"}},{"id":"meta/muse-spark-1.2","canonical_slug":"meta/muse-spark-1.2-20260805","hugging_face_id":null,"name":"Meta: Muse Spark 1.2","created":1785959287,"description":"Muse Spark 1.2 is a reasoning model from Meta, designed for complex agentic tasks. It accepts text, images, video, audio, and PDF documents, returns text, and offers a 1M-token context...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.00000425","web_search":"0.0025","input_cache_read":"0.00000015"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","repetition_penalty","response_format","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/meta/muse-spark-1.2-20260805/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1179,"win_rate":42.7,"rank":15},{"arena":"agents","category":"androidnative","elo":1236,"win_rate":53.3,"rank":10},{"arena":"agents","category":"fullstack","elo":1220,"win_rate":49.2,"rank":16},{"arena":"agents","category":"htmlslides","elo":1137,"win_rate":40,"rank":20},{"arena":"agents","category":"mobileapps","elo":1190,"win_rate":46.8,"rank":21},{"arena":"agents","category":"python-pptxslides","elo":1198,"win_rate":48.6,"rank":15},{"arena":"agents","category":"webapps","elo":1239,"win_rate":52.3,"rank":11},{"arena":"models","category":"3d","elo":1334,"win_rate":55.8,"rank":11},{"arena":"models","category":"asciiart","elo":1318,"win_rate":58.5,"rank":4},{"arena":"models","category":"codecategories","elo":1327,"win_rate":54.4,"rank":6},{"arena":"models","category":"dataviz","elo":1358,"win_rate":61.2,"rank":4},{"arena":"models","category":"gamedev","elo":1326,"win_rate":52.8,"rank":10},{"arena":"models","category":"svg","elo":1317,"win_rate":57.2,"rank":6},{"arena":"models","category":"uicomponent","elo":1326,"win_rate":54.1,"rank":10},{"arena":"models","category":"website","elo":1321,"win_rate":53.9,"rank":4}],"artificial_analysis":{"intelligence_index":39.8,"coding_index":72.2,"agentic_index":44}},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium","low","minimal"],"default_effort":"medium"}},{"id":"~deepseek/deepseek-v4-flash-latest","canonical_slug":"~deepseek/deepseek-v4-flash-latest","alias_target":{"name":"DeepSeek: DeepSeek V4 Flash 0731","slug":"deepseek/deepseek-v4-flash-0731"},"hugging_face_id":null,"name":"DeepSeek: DeepSeek V4 Flash Latest","created":1785606009,"description":"This model always redirects to the latest model in the DeepSeek V4 Flash family.","context_length":1310720,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.00000004","completion":"0.0000001","input_cache_read":"0.00000001"},"top_provider":{"context_length":1048576,"max_completion_tokens":393216,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","parallel_tool_calls","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_a","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":[],"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~deepseek/deepseek-v4-flash-latest/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"high"}},{"id":"deepseek/deepseek-v4-flash-0731","canonical_slug":"deepseek/deepseek-v4-flash-20260731","hugging_face_id":"deepseek-ai/DeepSeek-V4-Flash-0731","name":"DeepSeek: DeepSeek V4 Flash 0731","created":1785478908,"description":"DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows....","context_length":1310720,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.00000006","completion":"0.00000012","input_cache_read":"0.000000012"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","parallel_tool_calls","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_a","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":[],"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v4-flash-20260731/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1237,"win_rate":48.6,"rank":37},{"arena":"models","category":"asciiart","elo":1110,"win_rate":33.3,"rank":58},{"arena":"models","category":"codecategories","elo":1245,"win_rate":46.5,"rank":40},{"arena":"models","category":"dataviz","elo":1198,"win_rate":41.2,"rank":54},{"arena":"models","category":"gamedev","elo":1233,"win_rate":44.7,"rank":38},{"arena":"models","category":"svg","elo":1213,"win_rate":45.1,"rank":24},{"arena":"models","category":"uicomponent","elo":1253,"win_rate":47,"rank":35},{"arena":"models","category":"website","elo":1250,"win_rate":46.8,"rank":38}],"artificial_analysis":{"intelligence_index":34.5,"coding_index":69.1,"agentic_index":41.7}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"high"}},{"id":"deepseek/deepseek-v4-flash-0731:batch","canonical_slug":"deepseek/deepseek-v4-flash-20260731","hugging_face_id":"deepseek-ai/DeepSeek-V4-Flash-0731","name":"DeepSeek: DeepSeek V4 Flash 0731 (batch)","created":1785478908,"description":"DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows....","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.00000011","completion":"0.00000033","input_cache_read":"0.0000000035"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":[],"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v4-flash-20260731/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1237,"win_rate":48.6,"rank":37},{"arena":"models","category":"asciiart","elo":1110,"win_rate":33.3,"rank":58},{"arena":"models","category":"codecategories","elo":1245,"win_rate":46.5,"rank":40},{"arena":"models","category":"dataviz","elo":1198,"win_rate":41.2,"rank":54},{"arena":"models","category":"gamedev","elo":1233,"win_rate":44.7,"rank":38},{"arena":"models","category":"svg","elo":1213,"win_rate":45.1,"rank":24},{"arena":"models","category":"uicomponent","elo":1253,"win_rate":47,"rank":35},{"arena":"models","category":"website","elo":1250,"win_rate":46.8,"rank":38}],"artificial_analysis":{"intelligence_index":34.5,"coding_index":69.1,"agentic_index":41.7}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"high"}},{"id":"thinkingmachines/inkling-small","canonical_slug":"thinkingmachines/inkling-small-20260730","hugging_face_id":"thinkingmachines/Inkling-Small","name":"Thinking Machines: Inkling Small","created":1785443117,"description":"Inkling Small is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 12B active parameters out of 276B total. It is positioned as the smaller, more efficient member of...","context_length":1048576,"architecture":{"modality":"text+image+audio->text","input_modalities":["text","image","audio"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000045","completion":"0.0000012","input_cache_read":"0.0000001"},"top_provider":{"context_length":524288,"max_completion_tokens":262144,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","seed","stop","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/thinkingmachines/inkling-small-20260730/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":26.1,"coding_index":52.9,"agentic_index":25}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","medium","low","minimal","none"],"default_effort":"high"}},{"id":"thinkingmachines/inkling-small:batch","canonical_slug":"thinkingmachines/inkling-small-20260730","hugging_face_id":"thinkingmachines/Inkling-Small","name":"Thinking Machines: Inkling Small (batch)","created":1785443117,"description":"Inkling Small is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 12B active parameters out of 276B total. It is positioned as the smaller, more efficient member of...","context_length":524288,"architecture":{"modality":"text+image+audio->text","input_modalities":["text","image","audio"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000005","completion":"0.0000012","input_cache_read":"0.0000001"},"top_provider":{"context_length":524288,"max_completion_tokens":471859,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","stop","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/thinkingmachines/inkling-small-20260730/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":26.1,"coding_index":52.9,"agentic_index":25}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","medium","low","minimal","none"],"default_effort":"high"}},{"id":"thinkingmachines/inkling-small:free","canonical_slug":"thinkingmachines/inkling-small-20260730","hugging_face_id":"thinkingmachines/Inkling-Small","name":"Thinking Machines: Inkling Small (free)","created":1785443117,"description":"Inkling Small is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 12B active parameters out of 276B total. It is positioned as the smaller, more efficient member of...","context_length":1048576,"architecture":{"modality":"text+image+audio->text","input_modalities":["text","image","audio"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":1048576,"max_completion_tokens":262144,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","reasoning_effort","seed","stop","temperature","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/thinkingmachines/inkling-small-20260730/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":26.1,"coding_index":52.9,"agentic_index":25}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","medium","low","minimal","none"],"default_effort":"high"}},{"id":"qwen/qwen3.7-flash","canonical_slug":"qwen/qwen3.7-flash-20260727","hugging_face_id":null,"name":"Qwen: Qwen3.7 Flash","created":1785190561,"description":"Qwen3.7 Flash is a vision-language reasoning model from Alibaba. It is suited for multimodal agents, visual coding, search, and computer interaction, with strengths in object recognition, spatial understanding, and real-world...","context_length":1000000,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.00000003","completion":"0.00000013","input_cache_read":"0.000000006","input_cache_write":"0.000000038","overrides":[{"min_prompt_tokens":32000,"prompt":"0.0000001","completion":"0.0000004","input_cache_read":"0.00000002","input_cache_write":"0.000000125"},{"min_prompt_tokens":256000,"prompt":"0.0000002","completion":"0.0000008","input_cache_read":"0.00000004","input_cache_write":"0.00000025"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.7-flash-20260727/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supports_max_tokens":true}},{"id":"anthropic/claude-opus-5","canonical_slug":"anthropic/claude-opus-5-20260723","hugging_face_id":null,"name":"Claude Opus 5","created":1784912544,"description":"Claude Opus 5 is Anthropic’s flagship model for demanding reasoning, coding, and long-horizon agentic work. It is particularly strong at end-to-end software tasks, code review and bug finding, visual analysis...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000025","web_search":"0.01","input_cache_read":"0.0000005","input_cache_write":"0.00000625","input_cache_write_1h":"0.00001"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-opus-5-20260723/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1261,"win_rate":53.8,"rank":2},{"arena":"agents","category":"androidnative","elo":1265,"win_rate":54.9,"rank":4},{"arena":"agents","category":"fullstack","elo":1326,"win_rate":62,"rank":3},{"arena":"agents","category":"mobileapps","elo":1348,"win_rate":67.1,"rank":1},{"arena":"agents","category":"python-pptxslides","elo":1264,"win_rate":54,"rank":5},{"arena":"agents","category":"webapps","elo":1275,"win_rate":56.5,"rank":4},{"arena":"models","category":"3d","elo":1363,"win_rate":61.7,"rank":5},{"arena":"models","category":"asciiart","elo":1389,"win_rate":70.9,"rank":1},{"arena":"models","category":"codecategories","elo":1338,"win_rate":58.3,"rank":4},{"arena":"models","category":"dataviz","elo":1355,"win_rate":61.2,"rank":5},{"arena":"models","category":"gamedev","elo":1363,"win_rate":59.4,"rank":5},{"arena":"models","category":"svg","elo":1350,"win_rate":62,"rank":2},{"arena":"models","category":"uicomponent","elo":1361,"win_rate":61.4,"rank":3},{"arena":"models","category":"website","elo":1319,"win_rate":56.4,"rank":5}],"artificial_analysis":{"intelligence_index":50.7,"coding_index":78,"agentic_index":56.2}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"anthropic/claude-opus-5:batch","canonical_slug":"anthropic/claude-opus-5-20260723","hugging_face_id":null,"name":"Claude Opus 5 (batch)","created":1784912544,"description":"Claude Opus 5 is Anthropic’s flagship model for demanding reasoning, coding, and long-horizon agentic work. It is particularly strong at end-to-end software tasks, code review and bug finding, visual analysis...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.0000125","web_search":"0.01","input_cache_read":"0.00000025","input_cache_write":"0.000003125","input_cache_write_1h":"0.000005"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-opus-5-20260723/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1261,"win_rate":53.8,"rank":2},{"arena":"agents","category":"androidnative","elo":1265,"win_rate":54.9,"rank":4},{"arena":"agents","category":"fullstack","elo":1326,"win_rate":62,"rank":3},{"arena":"agents","category":"mobileapps","elo":1348,"win_rate":67.1,"rank":1},{"arena":"agents","category":"python-pptxslides","elo":1264,"win_rate":54,"rank":5},{"arena":"agents","category":"webapps","elo":1275,"win_rate":56.5,"rank":4},{"arena":"models","category":"3d","elo":1363,"win_rate":61.7,"rank":5},{"arena":"models","category":"asciiart","elo":1389,"win_rate":70.9,"rank":1},{"arena":"models","category":"codecategories","elo":1338,"win_rate":58.3,"rank":4},{"arena":"models","category":"dataviz","elo":1355,"win_rate":61.2,"rank":5},{"arena":"models","category":"gamedev","elo":1363,"win_rate":59.4,"rank":5},{"arena":"models","category":"svg","elo":1350,"win_rate":62,"rank":2},{"arena":"models","category":"uicomponent","elo":1361,"win_rate":61.4,"rank":3},{"arena":"models","category":"website","elo":1319,"win_rate":56.4,"rank":5}],"artificial_analysis":{"intelligence_index":50.7,"coding_index":78,"agentic_index":56.2}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"inclusionai/ling-3.0-flash","canonical_slug":"inclusionai/ling-3.0-flash-20260723","hugging_face_id":"inclusionAI/Ling-3.0-flash","name":"inclusionAI: Ling 3.0 Flash","created":1784818580,"description":"*Ling-3.0-flash* is a *124B-parameter Mixture-of-Experts (MoE) model*, with approximately *5.1B parameters activated per token*. The model is designed with *token efficiency and production-scale agentic inference* as key priorities, enabling developers...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000021","completion":"0.000000063","input_cache_read":"0.0000000042"},"top_provider":{"context_length":262144,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/inclusionai/ling-3.0-flash-20260723/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":50.6,"agentic_index":21}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"poolside/laguna-s-2.1","canonical_slug":"poolside/laguna-s-2.1-20260720","hugging_face_id":"poolside/Laguna-S-2.1","name":"Poolside: Laguna S 2.1","created":1784652683,"description":"Laguna S 2.1 is the latest coding agent model from [Poolside](). Laguna S 2.1 is a 118B total parameter model with 8B active parameters, scoring 70.2% on Terminal-Bench 2.1 and...","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000009","completion":"0.00000018","input_cache_read":"0.000000009"},"top_provider":{"context_length":1048576,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","temperature","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/poolside/laguna-s-2.1-20260720/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"poolside/laguna-s-2.1:free","canonical_slug":"poolside/laguna-s-2.1-20260720","hugging_face_id":"poolside/Laguna-S-2.1","name":"Poolside: Laguna S 2.1 (free)","created":1784652683,"description":"Laguna S 2.1 is the latest coding agent model from [Poolside](). Laguna S 2.1 is a 118B total parameter model with 8B active parameters, scoring 70.2% on Terminal-Bench 2.1 and...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":262144,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","temperature","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/poolside/laguna-s-2.1-20260720/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"google/gemini-3.6-flash","canonical_slug":"google/gemini-3.6-flash-20260721","hugging_face_id":null,"name":"Google: Gemini 3.6 Flash","created":1784646733,"description":"Gemini 3.6 Flash is a high-efficiency model from Google for coding, agentic workflows, and web and app development. It is designed to produce polished outputs with fewer unnecessary edits and...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000075","completion":"0.00000375","image":"0.00000075","audio":"0.00000075","input_audio_cache":"0.000000075","web_search":"0.014","internal_reasoning":"0.00000375","input_cache_read":"0.000000075","input_cache_write":"0.0000000416666666666667"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.6-flash-20260721/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1180,"win_rate":47.8,"rank":13},{"arena":"agents","category":"androidnative","elo":1213,"win_rate":53.3,"rank":12},{"arena":"agents","category":"fullstack","elo":1189,"win_rate":45.2,"rank":21},{"arena":"agents","category":"htmlslides","elo":1152,"win_rate":41.9,"rank":17},{"arena":"agents","category":"mobileapps","elo":1226,"win_rate":51.2,"rank":14},{"arena":"agents","category":"python-pptxslides","elo":1146,"win_rate":38.8,"rank":24},{"arena":"agents","category":"webapps","elo":1211,"win_rate":46.5,"rank":20},{"arena":"models","category":"3d","elo":1302,"win_rate":53.2,"rank":16},{"arena":"models","category":"asciiart","elo":1303,"win_rate":57,"rank":6},{"arena":"models","category":"codecategories","elo":1305,"win_rate":53.7,"rank":13},{"arena":"models","category":"dataviz","elo":1313,"win_rate":53.5,"rank":9},{"arena":"models","category":"gamedev","elo":1284,"win_rate":51,"rank":24},{"arena":"models","category":"uicomponent","elo":1318,"win_rate":55.3,"rank":11},{"arena":"models","category":"website","elo":1312,"win_rate":55.7,"rank":9}],"artificial_analysis":{"intelligence_index":34.3,"coding_index":69.2,"agentic_index":30.2}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"google/gemini-3.6-flash:batch","canonical_slug":"google/gemini-3.6-flash-20260721","hugging_face_id":null,"name":"Google: Gemini 3.6 Flash (batch)","created":1784646733,"description":"Gemini 3.6 Flash is a high-efficiency model from Google for coding, agentic workflows, and web and app development. It is designed to produce polished outputs with fewer unnecessary edits and...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.000000375","completion":"0.000001875","image":"0.000000375","audio":"0.000000375","input_audio_cache":"0.0000000375","web_search":"0.014","internal_reasoning":"0.000001875","input_cache_read":"0.0000000375","input_cache_write":"0.0000000416666666666667"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.6-flash-20260721/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1180,"win_rate":47.8,"rank":13},{"arena":"agents","category":"androidnative","elo":1213,"win_rate":53.3,"rank":12},{"arena":"agents","category":"fullstack","elo":1189,"win_rate":45.2,"rank":21},{"arena":"agents","category":"htmlslides","elo":1152,"win_rate":41.9,"rank":17},{"arena":"agents","category":"mobileapps","elo":1226,"win_rate":51.2,"rank":14},{"arena":"agents","category":"python-pptxslides","elo":1146,"win_rate":38.8,"rank":24},{"arena":"agents","category":"webapps","elo":1211,"win_rate":46.5,"rank":20},{"arena":"models","category":"3d","elo":1302,"win_rate":53.2,"rank":16},{"arena":"models","category":"asciiart","elo":1303,"win_rate":57,"rank":6},{"arena":"models","category":"codecategories","elo":1305,"win_rate":53.7,"rank":13},{"arena":"models","category":"dataviz","elo":1313,"win_rate":53.5,"rank":9},{"arena":"models","category":"gamedev","elo":1284,"win_rate":51,"rank":24},{"arena":"models","category":"uicomponent","elo":1318,"win_rate":55.3,"rank":11},{"arena":"models","category":"website","elo":1312,"win_rate":55.7,"rank":9}],"artificial_analysis":{"intelligence_index":34.3,"coding_index":69.2,"agentic_index":30.2}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"google/gemini-3.5-flash-lite","canonical_slug":"google/gemini-3.5-flash-lite-20260721","hugging_face_id":null,"name":"Google: Gemini 3.5 Flash Lite","created":1784646726,"description":"Gemini 3.5 Flash Lite is a high-efficiency model from Google with upgraded agentic capabilities. It is suited for subagents that execute focused tasks within complex, multi-agent workflows.","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000025","image":"0.0000003","audio":"0.0000003","input_audio_cache":"0.00000003","web_search":"0.014","internal_reasoning":"0.0000025","input_cache_read":"0.00000003","input_cache_write":"0.0000000833333333333333"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.5-flash-lite-20260721/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":22.7,"coding_index":49.3,"agentic_index":15.9}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"minimal"}},{"id":"google/gemini-3.5-flash-lite:batch","canonical_slug":"google/gemini-3.5-flash-lite-20260721","hugging_face_id":null,"name":"Google: Gemini 3.5 Flash Lite (batch)","created":1784646726,"description":"Gemini 3.5 Flash Lite is a high-efficiency model from Google with upgraded agentic capabilities. It is suited for subagents that execute focused tasks within complex, multi-agent workflows.","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.00000125","image":"0.00000015","audio":"0.00000015","input_audio_cache":"0.000000015","web_search":"0.014","internal_reasoning":"0.00000125","input_cache_read":"0.000000015"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.5-flash-lite-20260721/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":22.7,"coding_index":49.3,"agentic_index":15.9}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"minimal"}},{"id":"meituan/longcat-2.0","canonical_slug":"meituan/longcat-2.0-20260720","hugging_face_id":"meituan-longcat/LongCat-2.0","name":"Meituan: LongCat 2.0","created":1784554658,"description":"LongCat 2.0 is a sparse mixture-of-experts language model from Meituan, with 48B active parameters out of 1.6T total. It is suited for coding, repository-level changes, long-horizon problem solving, and agentic...","context_length":1048756,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000012","input_cache_read":"0.000000006"},"top_provider":{"context_length":1048756,"max_completion_tokens":262144,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/meituan/longcat-2.0-20260720/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":19.7,"coding_index":45.3,"agentic_index":15.9}},"reasoning":{"mandatory":false,"default_enabled":true,"supports_max_tokens":true}},{"id":"thinkingmachines/inkling","canonical_slug":"thinkingmachines/inkling-20260715","hugging_face_id":"thinkingmachines/Inkling","name":"Thinking Machines: Inkling","created":1784325956,"description":"Inkling is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 41B active parameters out of 975B total. It is designed for general-purpose reasoning, coding, agentic and tool-use systems,...","context_length":1048576,"architecture":{"modality":"text+image+audio->text","input_modalities":["text","image","audio"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.00000405","input_cache_read":"0.00000017"},"top_provider":{"context_length":524288,"max_completion_tokens":471859,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/thinkingmachines/inkling-20260715/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1187,"win_rate":41.4,"rank":53},{"arena":"models","category":"asciiart","elo":1118,"win_rate":34.6,"rank":55},{"arena":"models","category":"codecategories","elo":1215,"win_rate":42.1,"rank":48},{"arena":"models","category":"dataviz","elo":1193,"win_rate":40.4,"rank":57},{"arena":"models","category":"gamedev","elo":1178,"win_rate":37.2,"rank":59},{"arena":"models","category":"svg","elo":1135,"win_rate":34.9,"rank":52},{"arena":"models","category":"uicomponent","elo":1210,"win_rate":41.8,"rank":48},{"arena":"models","category":"website","elo":1227,"win_rate":43.5,"rank":47}],"artificial_analysis":{"intelligence_index":25.5,"coding_index":52.1,"agentic_index":24.3}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","medium","low","minimal","none"],"default_effort":"high"}},{"id":"thinkingmachines/inkling:batch","canonical_slug":"thinkingmachines/inkling-20260715","hugging_face_id":"thinkingmachines/Inkling","name":"Thinking Machines: Inkling (batch)","created":1784325956,"description":"Inkling is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 41B active parameters out of 975B total. It is designed for general-purpose reasoning, coding, agentic and tool-use systems,...","context_length":524288,"architecture":{"modality":"text+image+audio->text","input_modalities":["text","image","audio"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.00000405","input_cache_read":"0.00000017"},"top_provider":{"context_length":524288,"max_completion_tokens":471859,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/thinkingmachines/inkling-20260715/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1187,"win_rate":41.4,"rank":53},{"arena":"models","category":"asciiart","elo":1118,"win_rate":34.6,"rank":55},{"arena":"models","category":"codecategories","elo":1215,"win_rate":42.1,"rank":48},{"arena":"models","category":"dataviz","elo":1193,"win_rate":40.4,"rank":57},{"arena":"models","category":"gamedev","elo":1178,"win_rate":37.2,"rank":59},{"arena":"models","category":"svg","elo":1135,"win_rate":34.9,"rank":52},{"arena":"models","category":"uicomponent","elo":1210,"win_rate":41.8,"rank":48},{"arena":"models","category":"website","elo":1227,"win_rate":43.5,"rank":47}],"artificial_analysis":{"intelligence_index":25.5,"coding_index":52.1,"agentic_index":24.3}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","medium","low","minimal","none"],"default_effort":"high"}},{"id":"thinkingmachines/inkling:free","canonical_slug":"thinkingmachines/inkling-20260715","hugging_face_id":"thinkingmachines/Inkling","name":"Thinking Machines: Inkling (free)","created":1784325956,"description":"Inkling is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 41B active parameters out of 975B total. It is designed for general-purpose reasoning, coding, agentic and tool-use systems,...","context_length":1048576,"architecture":{"modality":"text+image+audio->text","input_modalities":["text","image","audio"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":1048576,"max_completion_tokens":262144,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","reasoning_effort","seed","stop","temperature","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/thinkingmachines/inkling-20260715/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1187,"win_rate":41.4,"rank":53},{"arena":"models","category":"asciiart","elo":1118,"win_rate":34.6,"rank":55},{"arena":"models","category":"codecategories","elo":1215,"win_rate":42.1,"rank":48},{"arena":"models","category":"dataviz","elo":1193,"win_rate":40.4,"rank":57},{"arena":"models","category":"gamedev","elo":1178,"win_rate":37.2,"rank":59},{"arena":"models","category":"svg","elo":1135,"win_rate":34.9,"rank":52},{"arena":"models","category":"uicomponent","elo":1210,"win_rate":41.8,"rank":48},{"arena":"models","category":"website","elo":1227,"win_rate":43.5,"rank":47}],"artificial_analysis":{"intelligence_index":25.5,"coding_index":52.1,"agentic_index":24.3}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","medium","low","minimal","none"],"default_effort":"high"}},{"id":"openrouter/auto-beta","canonical_slug":"openrouter/auto-beta","hugging_face_id":null,"name":"Auto Router (Beta)","created":1784311165,"description":"Auto Router (Beta) is a task-aware router from OpenRouter. It classifies each request, then routes it the [most popular model](/rankings#task-spend) for that task based on aggregate spend, filtered by your...","context_length":2000000,"architecture":{"modality":"text+image+file+audio+video->text+image","input_modalities":["text","image","audio","file","video"],"output_modalities":["text","image"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"-1","completion":"-1"},"top_provider":{"context_length":null,"max_completion_tokens":null,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","prediction","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_a","top_k","top_logprobs","top_p","web_search_options"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openrouter/auto-beta/endpoints"}},{"id":"moonshotai/kimi-k3","canonical_slug":"moonshotai/kimi-k3-20260715","hugging_face_id":"moonshotai/Kimi-K3","name":"MoonshotAI: Kimi K3","created":1784215858,"description":"Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI. It is suited for complex coding, knowledge work, and long-horizon agentic workflows, and is particularly strong at...","context_length":1048576,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000002648138063","completion":"0.00001328272425","input_cache_read":"0.00000030264435"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/moonshotai/kimi-k3-20260715/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1252,"win_rate":54.9,"rank":3},{"arena":"agents","category":"androidnative","elo":1255,"win_rate":54.7,"rank":6},{"arena":"agents","category":"fullstack","elo":1329,"win_rate":63.5,"rank":2},{"arena":"agents","category":"godotgamedev","elo":1199,"win_rate":48.5,"rank":10},{"arena":"agents","category":"htmlslides","elo":1259,"win_rate":59.6,"rank":2},{"arena":"agents","category":"mobileapps","elo":1274,"win_rate":57.5,"rank":4},{"arena":"agents","category":"python-pptxslides","elo":1273,"win_rate":59.2,"rank":4},{"arena":"agents","category":"webapps","elo":1312,"win_rate":61.2,"rank":2},{"arena":"models","category":"3d","elo":1426,"win_rate":69,"rank":2},{"arena":"models","category":"codecategories","elo":1388,"win_rate":64.7,"rank":1},{"arena":"models","category":"dataviz","elo":1365,"win_rate":64.1,"rank":2},{"arena":"models","category":"gamedev","elo":1404,"win_rate":63.1,"rank":2},{"arena":"models","category":"svg","elo":1336,"win_rate":63.1,"rank":3},{"arena":"models","category":"uicomponent","elo":1369,"win_rate":63,"rank":2},{"arena":"models","category":"website","elo":1352,"win_rate":60.8,"rank":2}],"artificial_analysis":{"intelligence_index":43.8,"coding_index":76.2,"agentic_index":50.6}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"max"}},{"id":"moonshotai/kimi-k3:batch","canonical_slug":"moonshotai/kimi-k3-20260715","hugging_face_id":"moonshotai/Kimi-K3","name":"MoonshotAI: Kimi K3 (batch)","created":1784215858,"description":"Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI. It is suited for complex coding, knowledge work, and long-horizon agentic workflows, and is particularly strong at...","context_length":1048576,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000003","completion":"0.000015","input_cache_read":"0.0000003"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/moonshotai/kimi-k3-20260715/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1252,"win_rate":54.9,"rank":3},{"arena":"agents","category":"androidnative","elo":1255,"win_rate":54.7,"rank":6},{"arena":"agents","category":"fullstack","elo":1329,"win_rate":63.5,"rank":2},{"arena":"agents","category":"godotgamedev","elo":1199,"win_rate":48.5,"rank":10},{"arena":"agents","category":"htmlslides","elo":1259,"win_rate":59.6,"rank":2},{"arena":"agents","category":"mobileapps","elo":1274,"win_rate":57.5,"rank":4},{"arena":"agents","category":"python-pptxslides","elo":1273,"win_rate":59.2,"rank":4},{"arena":"agents","category":"webapps","elo":1312,"win_rate":61.2,"rank":2},{"arena":"models","category":"3d","elo":1426,"win_rate":69,"rank":2},{"arena":"models","category":"codecategories","elo":1388,"win_rate":64.7,"rank":1},{"arena":"models","category":"dataviz","elo":1365,"win_rate":64.1,"rank":2},{"arena":"models","category":"gamedev","elo":1404,"win_rate":63.1,"rank":2},{"arena":"models","category":"svg","elo":1336,"win_rate":63.1,"rank":3},{"arena":"models","category":"uicomponent","elo":1369,"win_rate":63,"rank":2},{"arena":"models","category":"website","elo":1352,"win_rate":60.8,"rank":2}],"artificial_analysis":{"intelligence_index":43.8,"coding_index":76.2,"agentic_index":50.6}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"max"}},{"id":"meta/muse-spark-1.1","canonical_slug":"meta/muse-spark-1.1-20260709","hugging_face_id":null,"name":"Meta: Muse Spark 1.1","created":1784215741,"description":"Muse Spark 1.1 is a multimodal reasoning model from Meta, built for agentic tasks. It accepts text, images, video, audio, and PDF documents and returns text, with a 1M-token context...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.00000425","web_search":"0.0025","input_cache_read":"0.00000015"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","repetition_penalty","response_format","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/meta/muse-spark-1.1-20260709/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1183,"win_rate":48.3,"rank":11},{"arena":"agents","category":"androidnative","elo":1188,"win_rate":47.5,"rank":18},{"arena":"agents","category":"fullstack","elo":1213,"win_rate":47.4,"rank":17},{"arena":"agents","category":"godotgamedev","elo":1141,"win_rate":39.4,"rank":19},{"arena":"agents","category":"htmlslides","elo":1197,"win_rate":51.8,"rank":8},{"arena":"agents","category":"mobileapps","elo":1198,"win_rate":46.8,"rank":16},{"arena":"agents","category":"python-pptxslides","elo":1156,"win_rate":44.1,"rank":21},{"arena":"agents","category":"webapps","elo":1215,"win_rate":49.4,"rank":18},{"arena":"models","category":"3d","elo":1294,"win_rate":52.1,"rank":19},{"arena":"models","category":"asciiart","elo":1307,"win_rate":58.5,"rank":5},{"arena":"models","category":"codecategories","elo":1291,"win_rate":52.1,"rank":19},{"arena":"models","category":"dataviz","elo":1289,"win_rate":52.1,"rank":16},{"arena":"models","category":"gamedev","elo":1297,"win_rate":49.3,"rank":17},{"arena":"models","category":"svg","elo":1278,"win_rate":52.3,"rank":10},{"arena":"models","category":"uicomponent","elo":1310,"win_rate":52.7,"rank":14},{"arena":"models","category":"website","elo":1281,"win_rate":52,"rank":23}],"artificial_analysis":{"intelligence_index":34.3,"coding_index":71.3,"agentic_index":27.5}},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium","low","minimal"],"default_effort":"medium"}},{"id":"kwaipilot/kat-coder-pro-v2.5","canonical_slug":"kwaipilot/kat-coder-pro-v2.5-20260710","hugging_face_id":null,"name":"Kwaipilot: KAT-Coder-Pro V2.5","created":1783714589,"description":"KAT-Coder-Pro V2.5 is a flagship-level Agentic Coding model that can directly hand over an entire issue or an entire business workflow to it, allowing it to autonomously locate and make...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000074","completion":"0.00000296","input_cache_read":"0.00000015"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/kwaipilot/kat-coder-pro-v2.5-20260710/endpoints"}},{"id":"openai/gpt-5.6-luna-pro","canonical_slug":"openai/gpt-5.6-luna-pro-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Luna Pro","created":1783590867,"description":"GPT-5.6 Luna Pro is the same underlying model as [GPT-5.6 Luna](https://openrouter.ai/openai/gpt-5.6-luna), served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks.\n\nLearn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.0000012","web_search":"0.01","input_cache_read":"0.00000002","input_cache_write":"0.00000025","overrides":[{"min_prompt_tokens":272000,"prompt":"0.0000004","completion":"0.0000018","input_cache_read":"0.00000004","input_cache_write":"0.0000005"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-luna-pro-20260709/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.6-luna-pro:batch","canonical_slug":"openai/gpt-5.6-luna-pro-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Luna Pro (batch)","created":1783590867,"description":"GPT-5.6 Luna Pro is the same underlying model as [GPT-5.6 Luna](https://openrouter.ai/openai/gpt-5.6-luna), served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks.\n\nLearn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000006","web_search":"0.01","input_cache_read":"0.00000001","overrides":[{"min_prompt_tokens":272000,"prompt":"0.0000002","completion":"0.0000009","input_cache_read":"0.00000002"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-luna-pro-20260709/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.6-luna","canonical_slug":"openai/gpt-5.6-luna-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Luna","created":1783590864,"description":"GPT-5.6 Luna is a fast, cost-efficient model in OpenAI's GPT-5.6 series. It is suited for high-volume, latency-sensitive tasks such as chat, classification, and lightweight agentic workflows, providing capable reasoning for...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.0000012","web_search":"0.01","input_cache_read":"0.00000002","input_cache_write":"0.00000025","overrides":[{"min_prompt_tokens":272000,"prompt":"0.0000004","completion":"0.0000018","input_cache_read":"0.00000004","input_cache_write":"0.0000005"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-luna-20260709/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":37.5,"coding_index":71.4,"agentic_index":42.7}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.6-luna:batch","canonical_slug":"openai/gpt-5.6-luna-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Luna (batch)","created":1783590864,"description":"GPT-5.6 Luna is a fast, cost-efficient model in OpenAI's GPT-5.6 series. It is suited for high-volume, latency-sensitive tasks such as chat, classification, and lightweight agentic workflows, providing capable reasoning for...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000006","web_search":"0.01","input_cache_read":"0.00000001","overrides":[{"min_prompt_tokens":272000,"prompt":"0.0000002","completion":"0.0000009","input_cache_read":"0.00000002"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-luna-20260709/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":37.5,"coding_index":71.4,"agentic_index":42.7}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.6-terra-pro","canonical_slug":"openai/gpt-5.6-terra-pro-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Terra Pro","created":1783590861,"description":"GPT-5.6 Terra Pro is the same underlying model as [GPT-5.6 Terra](https://openrouter.ai/openai/gpt-5.6-terra), served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks.\n\nLearn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000012","web_search":"0.01","input_cache_read":"0.0000002","input_cache_write":"0.0000025","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000004","completion":"0.000018","input_cache_read":"0.0000004","input_cache_write":"0.000005"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-terra-pro-20260709/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.6-terra-pro:batch","canonical_slug":"openai/gpt-5.6-terra-pro-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Terra Pro (batch)","created":1783590861,"description":"GPT-5.6 Terra Pro is the same underlying model as [GPT-5.6 Terra](https://openrouter.ai/openai/gpt-5.6-terra), served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks.\n\nLearn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000006","web_search":"0.01","input_cache_read":"0.0000001","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000002","completion":"0.000009","input_cache_read":"0.0000002"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-terra-pro-20260709/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.6-terra","canonical_slug":"openai/gpt-5.6-terra-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Terra","created":1783590857,"description":"GPT-5.6 Terra is a balanced model in OpenAI's GPT-5.6 series, positioned between the flagship Sol tier and the cost-efficient Luna tier. It is suited for everyday coding, reasoning, and agentic...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000012","web_search":"0.01","input_cache_read":"0.0000002","input_cache_write":"0.0000025","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000004","completion":"0.000018","input_cache_read":"0.0000004","input_cache_write":"0.000005"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-terra-20260709/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":42.3,"coding_index":76.7,"agentic_index":43.7}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.6-terra:batch","canonical_slug":"openai/gpt-5.6-terra-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Terra (batch)","created":1783590857,"description":"GPT-5.6 Terra is a balanced model in OpenAI's GPT-5.6 series, positioned between the flagship Sol tier and the cost-efficient Luna tier. It is suited for everyday coding, reasoning, and agentic...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000006","web_search":"0.01","input_cache_read":"0.0000001","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000002","completion":"0.000009","input_cache_read":"0.0000002"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-terra-20260709/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":42.3,"coding_index":76.7,"agentic_index":43.7}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.6-sol-pro","canonical_slug":"openai/gpt-5.6-sol-pro-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Sol Pro","created":1783590854,"description":"GPT-5.6 Sol Pro is the same underlying model as [GPT-5.6 Sol](https://openrouter.ai/openai/gpt-5.6-sol), served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks.\n\nLearn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.00001","web_search":"0.01","input_cache_read":"0.0000002","input_cache_write":"0.0000025","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000004","completion":"0.000015","input_cache_read":"0.0000004","input_cache_write":"0.000005"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-sol-pro-20260709/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.6-sol-pro:batch","canonical_slug":"openai/gpt-5.6-sol-pro-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Sol Pro (batch)","created":1783590854,"description":"GPT-5.6 Sol Pro is the same underlying model as [GPT-5.6 Sol](https://openrouter.ai/openai/gpt-5.6-sol), served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks.\n\nLearn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000005","web_search":"0.01","input_cache_read":"0.0000001","input_cache_write":"0.00000125","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000002","completion":"0.0000075","input_cache_read":"0.0000002","input_cache_write":"0.0000025"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-sol-pro-20260709/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.6-sol","canonical_slug":"openai/gpt-5.6-sol-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Sol","created":1783590850,"description":"GPT-5.6 Sol is the flagship model in OpenAI's GPT-5.6 series. It is suited for complex reasoning, coding, and agentic workflows, and is particularly strong at command-line and multi-step coding tasks...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.00001","web_search":"0.01","input_cache_read":"0.0000002","input_cache_write":"0.0000025","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000004","completion":"0.000015","input_cache_read":"0.0000004","input_cache_write":"0.000005"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-sol-20260709/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":47.1,"coding_index":77.4,"agentic_index":50.5}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.6-sol:batch","canonical_slug":"openai/gpt-5.6-sol-20260709","hugging_face_id":null,"name":"OpenAI: GPT-5.6 Sol (batch)","created":1783590850,"description":"GPT-5.6 Sol is the flagship model in OpenAI's GPT-5.6 series. It is suited for complex reasoning, coding, and agentic workflows, and is particularly strong at command-line and multi-step coding tasks...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000005","web_search":"0.01","input_cache_read":"0.0000001","input_cache_write":"0.00000125","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000002","completion":"0.0000075","input_cache_read":"0.0000002","input_cache_write":"0.0000025"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2026-02-16","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.6-sol-20260709/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":47.1,"coding_index":77.4,"agentic_index":50.5}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"x-ai/grok-4.5","canonical_slug":"x-ai/grok-4.5-20260708","hugging_face_id":null,"name":"SpaceXAI: Grok 4.5","created":1783523154,"description":"Grok 4.5 is a model from SpaceXAI with frontier performance on coding, knowledge work, and STEM.","context_length":500000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Grok","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000006","web_search":"0.005","input_cache_read":"0.0000003","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000004","completion":"0.000012","input_cache_read":"0.0000006"}]},"top_provider":{"context_length":500000,"max_completion_tokens":450000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","logprobs","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/x-ai/grok-4.5-20260708/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1235,"win_rate":55.9,"rank":6},{"arena":"agents","category":"androidnative","elo":1247,"win_rate":64.1,"rank":8},{"arena":"agents","category":"fullstack","elo":1260,"win_rate":62.1,"rank":8},{"arena":"agents","category":"godotgamedev","elo":1270,"win_rate":62.1,"rank":3},{"arena":"agents","category":"htmlslides","elo":1208,"win_rate":52.5,"rank":6},{"arena":"agents","category":"mobileapps","elo":1248,"win_rate":54.6,"rank":10},{"arena":"agents","category":"python-pptxslides","elo":1230,"win_rate":53.9,"rank":12},{"arena":"agents","category":"webapps","elo":1229,"win_rate":53.9,"rank":14},{"arena":"models","category":"3d","elo":1289,"win_rate":50.6,"rank":20},{"arena":"models","category":"asciiart","elo":1269,"win_rate":54.5,"rank":14},{"arena":"models","category":"codecategories","elo":1292,"win_rate":51.3,"rank":17},{"arena":"models","category":"dataviz","elo":1287,"win_rate":49.8,"rank":17},{"arena":"models","category":"gamedev","elo":1288,"win_rate":49.4,"rank":19},{"arena":"models","category":"svg","elo":1250,"win_rate":50.1,"rank":14},{"arena":"models","category":"uicomponent","elo":1294,"win_rate":52.1,"rank":19},{"arena":"models","category":"website","elo":1296,"win_rate":53.2,"rank":15}],"artificial_analysis":{"intelligence_index":39.1,"coding_index":72.4,"agentic_index":42.1}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low"],"default_effort":"high"}},{"id":"~x-ai/grok-latest","canonical_slug":"~x-ai/grok-latest","alias_target":{"name":"SpaceXAI: Grok 4.6","slug":"x-ai/grok-4.6"},"hugging_face_id":null,"name":"xAI: Grok Latest","created":1783519360,"description":"This model always redirects to the latest Grok model from xAI.","context_length":500000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000006","web_search":"0.005","input_cache_read":"0.0000005","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000004","completion":"0.000012","input_cache_read":"0.000001"}]},"top_provider":{"context_length":500000,"max_completion_tokens":450000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","logprobs","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~x-ai/grok-latest/endpoints"},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["xhigh","high","medium","low"],"default_effort":"high"}},{"id":"aion-labs/aion-3.0-mini","canonical_slug":"aion-labs/aion-3.0-mini-20260707","hugging_face_id":null,"name":"AionLabs: Aion-3.0-Mini","created":1783443096,"description":"Aion-3.0 Mini is a multi-model roleplaying and storytelling system from AionLabs, built on the DeepSeek family of models. It uses a collaborative generation process in which multiple specialized models each...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000007","completion":"0.0000014","input_cache_read":"0.00000018"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/aion-labs/aion-3.0-mini-20260707/endpoints"},"reasoning":{"mandatory":true}},{"id":"aion-labs/aion-3.0","canonical_slug":"aion-labs/aion-3.0-20260707","hugging_face_id":null,"name":"AionLabs: Aion-3.0","created":1783443095,"description":"Aion-3.0 is a multi-model roleplaying and storytelling system from AionLabs, built on the GLM family of models. It uses a collaborative generation process in which multiple specialized models each contribute...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000003","completion":"0.000006","input_cache_read":"0.00000075"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/aion-labs/aion-3.0-20260707/endpoints"},"reasoning":{"mandatory":true}},{"id":"tencent/hy3","canonical_slug":"tencent/hy3-20260706","hugging_face_id":"tencent/Hy3","name":"Tencent: Hy3","created":1783344048,"description":"Hy3 is a 295B-parameter Mixture-of-Experts model from Tencent (21B active, 192 experts with top-8 routing) built for reasoning, agentic workflows, and real-world production use. It supports a configurable reasoning effort:...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000000825","completion":"0.00000033","input_cache_read":"0.000000020625","overrides":[{"utc_start":0,"utc_end":1600,"prompt":"0.000000132","completion":"0.000000528","input_cache_read":"0.000000033"},{"utc_start":1600,"utc_end":0,"prompt":"0.0000000825","completion":"0.00000033","input_cache_read":"0.000000020625"}]},"top_provider":{"context_length":262144,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_completion_tokens","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":0.9,"top_p":1,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/tencent/hy3-20260706/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1205,"win_rate":43.8,"rank":47},{"arena":"models","category":"codecategories","elo":1191,"win_rate":41.1,"rank":58},{"arena":"models","category":"dataviz","elo":1142,"win_rate":36.1,"rank":81},{"arena":"models","category":"gamedev","elo":1159,"win_rate":38.6,"rank":67},{"arena":"models","category":"uicomponent","elo":1179,"win_rate":40.2,"rank":63},{"arena":"models","category":"website","elo":1194,"win_rate":41.3,"rank":62}]},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","low","none"],"default_effort":"high"}},{"id":"poolside/laguna-xs-2.1","canonical_slug":"poolside/laguna-xs-2.1-20260625","hugging_face_id":"poolside/Laguna-XS-2.1","name":"Poolside: Laguna XS 2.1","created":1783002429,"description":"Laguna XS 2.1 is the latest coding agent model in the 33B-A3B category from [Poolside](https://poolside.ai/) and a step forward from their Laguna XS.2 model (released in April 2026). It combines...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000006","completion":"0.00000012","input_cache_read":"0.00000003"},"top_provider":{"context_length":262144,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","temperature","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/poolside/laguna-xs-2.1-20260625/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"poolside/laguna-xs-2.1:free","canonical_slug":"poolside/laguna-xs-2.1-20260625","hugging_face_id":"poolside/Laguna-XS-2.1","name":"Poolside: Laguna XS 2.1 (free)","created":1783002429,"description":"Laguna XS 2.1 is the latest coding agent model in the 33B-A3B category from [Poolside](https://poolside.ai/) and a step forward from their Laguna XS.2 model (released in April 2026). It combines...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":262144,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","temperature","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/poolside/laguna-xs-2.1-20260625/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"anthropic/claude-sonnet-5","canonical_slug":"anthropic/claude-sonnet-5-20260630","hugging_face_id":null,"name":"Anthropic: Claude Sonnet 5","created":1782843083,"description":"Sonnet 5 is Anthropic's most capable Sonnet-class model, with frontier performance across coding, agents, and professional work. It supports adaptive thinking with selectable reasoning effort levels (low, medium, high, max,...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.00001","web_search":"0.01","input_cache_read":"0.0000002","input_cache_write":"0.0000025","input_cache_write_1h":"0.000004"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-sonnet-5-20260630/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1226,"win_rate":52.3,"rank":7},{"arena":"agents","category":"androidnative","elo":1244,"win_rate":55.1,"rank":9},{"arena":"agents","category":"fullstack","elo":1259,"win_rate":55.3,"rank":9},{"arena":"agents","category":"godotgamedev","elo":1270,"win_rate":60.6,"rank":2},{"arena":"agents","category":"htmlslides","elo":1220,"win_rate":53.7,"rank":5},{"arena":"agents","category":"mobileapps","elo":1255,"win_rate":55.7,"rank":7},{"arena":"agents","category":"python-pptxslides","elo":1216,"win_rate":50.1,"rank":13},{"arena":"agents","category":"webapps","elo":1258,"win_rate":54.9,"rank":8},{"arena":"models","category":"3d","elo":1288,"win_rate":54.8,"rank":21},{"arena":"models","category":"asciiart","elo":1226,"win_rate":52.1,"rank":19},{"arena":"models","category":"codecategories","elo":1292,"win_rate":53.8,"rank":16},{"arena":"models","category":"dataviz","elo":1260,"win_rate":52.8,"rank":26},{"arena":"models","category":"gamedev","elo":1313,"win_rate":54,"rank":13},{"arena":"models","category":"svg","elo":1215,"win_rate":51.6,"rank":23},{"arena":"models","category":"uicomponent","elo":1297,"win_rate":55.1,"rank":17},{"arena":"models","category":"website","elo":1289,"win_rate":53.6,"rank":18}],"artificial_analysis":{"intelligence_index":38.4,"coding_index":71.5,"agentic_index":44.3}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"anthropic/claude-sonnet-5:batch","canonical_slug":"anthropic/claude-sonnet-5-20260630","hugging_face_id":null,"name":"Anthropic: Claude Sonnet 5 (batch)","created":1782843083,"description":"Sonnet 5 is Anthropic's most capable Sonnet-class model, with frontier performance across coding, agents, and professional work. It supports adaptive thinking with selectable reasoning effort levels (low, medium, high, max,...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000005","web_search":"0.01","input_cache_read":"0.0000001","input_cache_write":"0.00000125","input_cache_write_1h":"0.000002"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-sonnet-5-20260630/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1226,"win_rate":52.3,"rank":7},{"arena":"agents","category":"androidnative","elo":1244,"win_rate":55.1,"rank":9},{"arena":"agents","category":"fullstack","elo":1259,"win_rate":55.3,"rank":9},{"arena":"agents","category":"godotgamedev","elo":1270,"win_rate":60.6,"rank":2},{"arena":"agents","category":"htmlslides","elo":1220,"win_rate":53.7,"rank":5},{"arena":"agents","category":"mobileapps","elo":1255,"win_rate":55.7,"rank":7},{"arena":"agents","category":"python-pptxslides","elo":1216,"win_rate":50.1,"rank":13},{"arena":"agents","category":"webapps","elo":1258,"win_rate":54.9,"rank":8},{"arena":"models","category":"3d","elo":1288,"win_rate":54.8,"rank":21},{"arena":"models","category":"asciiart","elo":1226,"win_rate":52.1,"rank":19},{"arena":"models","category":"codecategories","elo":1292,"win_rate":53.8,"rank":16},{"arena":"models","category":"dataviz","elo":1260,"win_rate":52.8,"rank":26},{"arena":"models","category":"gamedev","elo":1313,"win_rate":54,"rank":13},{"arena":"models","category":"svg","elo":1215,"win_rate":51.6,"rank":23},{"arena":"models","category":"uicomponent","elo":1297,"win_rate":55.1,"rank":17},{"arena":"models","category":"website","elo":1289,"win_rate":53.6,"rank":18}],"artificial_analysis":{"intelligence_index":38.4,"coding_index":71.5,"agentic_index":44.3}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"google/gemini-3.1-flash-lite-image","canonical_slug":"google/gemini-3.1-flash-lite-image-20260630","hugging_face_id":null,"name":"Google: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)","created":1782837225,"description":"Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) is Google's fastest, most cost-efficient Gemini image model, built for high-velocity developer pipelines and rapid-fire visual exploration. It delivers text-to-image generation...","context_length":65536,"architecture":{"modality":"text+image->text+image","input_modalities":["image","text"],"output_modalities":["image","text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.0000015","image_output":"0.00003","web_search":"0.014"},"top_provider":{"context_length":65536,"max_completion_tokens":58982,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","temperature","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-01-01","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.1-flash-lite-image-20260630/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","minimal"],"default_effort":"minimal"}},{"id":"sakana/fugu-ultra","canonical_slug":"sakana/fugu-ultra-20260615","hugging_face_id":null,"name":"Sakana: Fugu Ultra","created":1782276303,"description":"Fugu Ultra is the higher-performance model in Sakana AI's Fugu family. Rather than a single monolithic model, Fugu is a learned multi-agent orchestration system: a language model trained to route...","context_length":1000000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.00003","web_search":"0.01","input_cache_read":"0.0000005","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00001","completion":"0.000045","input_cache_read":"0.000001"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","reasoning","reasoning_effort","structured_outputs","tool_choice","tools","web_search_options"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/sakana/fugu-ultra-20260615/endpoints"},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["max","xhigh","high"],"default_effort":"xhigh"}},{"id":"google/gemini-3.1-flash-image","canonical_slug":"google/gemini-3.1-flash-image-20260528","hugging_face_id":null,"name":"Google: Nano Banana 2 (Gemini 3.1 Flash Image)","created":1781754065,"description":"Gemini 3.1 Flash Image, a.k.a. \"Nano Banana 2,\" is Google’s latest state of the art image generation and editing model, delivering Pro-level visual quality at Flash speed. It combines advanced...","context_length":131072,"architecture":{"modality":"text+image->text+image","input_modalities":["image","text"],"output_modalities":["image","text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.0000005","completion":"0.000003","image_output":"0.00006","web_search":"0.014"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","temperature","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.1-flash-image-20260528/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","minimal"],"default_effort":"minimal"}},{"id":"google/gemini-3-pro-image","canonical_slug":"google/gemini-3-pro-image-20260528","hugging_face_id":null,"name":"Google: Nano Banana Pro (Gemini 3 Pro Image)","created":1781754054,"description":"Nano Banana Pro is Google’s most advanced image-generation and editing model, built on Gemini 3 Pro. It extends the original Nano Banana with significantly improved multimodal reasoning, real-world grounding, and...","context_length":131072,"architecture":{"modality":"text+image->text+image","input_modalities":["image","text"],"output_modalities":["image","text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000012","image":"0.000002","image_output":"0.00012","audio":"0.000002","input_audio_cache":"0.0000002","web_search":"0.014","internal_reasoning":"0.000012","input_cache_read":"0.0000002","input_cache_write":"0.000000375"},"top_provider":{"context_length":65536,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3-pro-image-20260528/endpoints"},"reasoning":{"mandatory":true}},{"id":"cohere/north-mini-code:free","canonical_slug":"cohere/north-mini-code-20260617","hugging_face_id":"CohereLabs/North-Mini-Code-1.0","name":"Cohere: North Mini Code (free)","created":1781723748,"description":"North Mini Code is Cohere's first agentic coding model and the debut of its North family. A sparse mixture-of-experts model with 30B total parameters and 3B active, it is optimized...","context_length":256000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Cohere","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":256000,"max_completion_tokens":64000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/cohere/north-mini-code-20260617/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":36.5,"agentic_index":1.1}},"reasoning":{"mandatory":false}},{"id":"z-ai/glm-5.2","canonical_slug":"z-ai/glm-5.2-20260616","hugging_face_id":"zai-org/GLM-5.2","name":"Z.ai: GLM 5.2","created":1781631930,"description":"GLM 5.2 is a large-scale reasoning model from Z.ai. It supports text input and output with a 1M-token context window, and is suited for long-horizon agent workflows, project-level software engineering,...","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000006832","completion":"0.0000021472","input_cache_read":"0.00000012688"},"top_provider":{"context_length":1048576,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","parallel_tool_calls","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-5.2-20260616/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1205,"win_rate":49.1,"rank":10},{"arena":"agents","category":"androidnative","elo":1190,"win_rate":53,"rank":17},{"arena":"agents","category":"fullstack","elo":1249,"win_rate":61.2,"rank":11},{"arena":"agents","category":"godotgamedev","elo":1142,"win_rate":40.1,"rank":16},{"arena":"agents","category":"htmlslides","elo":1178,"win_rate":49.6,"rank":13},{"arena":"agents","category":"mobileapps","elo":1188,"win_rate":51,"rank":22},{"arena":"agents","category":"python-pptxslides","elo":1194,"win_rate":48.7,"rank":16},{"arena":"agents","category":"webapps","elo":1234,"win_rate":56.3,"rank":13},{"arena":"models","category":"3d","elo":1323,"win_rate":54.6,"rank":13},{"arena":"models","category":"asciiart","elo":1234,"win_rate":49.9,"rank":17},{"arena":"models","category":"codecategories","elo":1309,"win_rate":53.3,"rank":10},{"arena":"models","category":"dataviz","elo":1313,"win_rate":54,"rank":10},{"arena":"models","category":"gamedev","elo":1298,"win_rate":50.5,"rank":16},{"arena":"models","category":"svg","elo":1249,"win_rate":51.7,"rank":15},{"arena":"models","category":"uicomponent","elo":1309,"win_rate":55.7,"rank":15},{"arena":"models","category":"website","elo":1304,"win_rate":53.5,"rank":12}],"artificial_analysis":{"intelligence_index":null,"coding_index":68.8,"agentic_index":39.4}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["xhigh","high"],"default_effort":"high"}},{"id":"z-ai/glm-5.2:batch","canonical_slug":"z-ai/glm-5.2-20260616","hugging_face_id":"zai-org/GLM-5.2","name":"Z.ai: GLM 5.2 (batch)","created":1781631930,"description":"GLM 5.2 is a large-scale reasoning model from Z.ai. It supports text input and output with a 1M-token context window, and is suited for long-horizon agent workflows, project-level software engineering,...","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000007","completion":"0.0000022","input_cache_read":"0.00000007"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-5.2-20260616/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1205,"win_rate":49.1,"rank":10},{"arena":"agents","category":"androidnative","elo":1190,"win_rate":53,"rank":17},{"arena":"agents","category":"fullstack","elo":1249,"win_rate":61.2,"rank":11},{"arena":"agents","category":"godotgamedev","elo":1142,"win_rate":40.1,"rank":16},{"arena":"agents","category":"htmlslides","elo":1178,"win_rate":49.6,"rank":13},{"arena":"agents","category":"mobileapps","elo":1188,"win_rate":51,"rank":22},{"arena":"agents","category":"python-pptxslides","elo":1194,"win_rate":48.7,"rank":16},{"arena":"agents","category":"webapps","elo":1234,"win_rate":56.3,"rank":13},{"arena":"models","category":"3d","elo":1323,"win_rate":54.6,"rank":13},{"arena":"models","category":"asciiart","elo":1234,"win_rate":49.9,"rank":17},{"arena":"models","category":"codecategories","elo":1309,"win_rate":53.3,"rank":10},{"arena":"models","category":"dataviz","elo":1313,"win_rate":54,"rank":10},{"arena":"models","category":"gamedev","elo":1298,"win_rate":50.5,"rank":16},{"arena":"models","category":"svg","elo":1249,"win_rate":51.7,"rank":15},{"arena":"models","category":"uicomponent","elo":1309,"win_rate":55.7,"rank":15},{"arena":"models","category":"website","elo":1304,"win_rate":53.5,"rank":12}],"artificial_analysis":{"intelligence_index":null,"coding_index":68.8,"agentic_index":39.4}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["xhigh","high"],"default_effort":"high"}},{"id":"openrouter/fusion","canonical_slug":"openrouter/fusion","hugging_face_id":null,"name":"OpenRouter: Fusion","created":1781371647,"description":"Fusion turns your prompt into a small multi-model deliberation. A panel of expert models (see below) analyzes your prompt in parallel with web search and web fetch enabled, then a...","context_length":1000000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"-1","completion":"-1"},"top_provider":{"context_length":null,"max_completion_tokens":null,"is_moderated":false},"per_request_limits":null,"supported_parameters":[],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openrouter/fusion/endpoints"}},{"id":"moonshotai/kimi-k2.7-code","canonical_slug":"moonshotai/kimi-k2.7-code-20260612","hugging_face_id":"moonshotai/Kimi-K2.7-Code","name":"MoonshotAI: Kimi K2.7 Code","created":1781266361,"description":"MoonshotAI: Kimi K2.7 Code is a coding-focused model in Moonshot AI's Kimi K2 family, built to complete end-to-end programming tasks reliably over long contexts. It uses a native multimodal mixture-of-experts...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000071","completion":"0.0000035","input_cache_read":"0.00000015"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","parallel_tool_calls","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/moonshotai/kimi-k2.7-code-20260612/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1130,"win_rate":43,"rank":21},{"arena":"agents","category":"androidnative","elo":1151,"win_rate":46.1,"rank":25},{"arena":"agents","category":"fullstack","elo":1184,"win_rate":51.7,"rank":22},{"arena":"agents","category":"godotgamedev","elo":1185,"win_rate":49.5,"rank":11},{"arena":"agents","category":"htmlslides","elo":1204,"win_rate":53.6,"rank":7},{"arena":"agents","category":"mobileapps","elo":1171,"win_rate":47.5,"rank":26},{"arena":"agents","category":"python-pptxslides","elo":1154,"win_rate":44.7,"rank":22},{"arena":"agents","category":"webapps","elo":1189,"win_rate":47.3,"rank":22},{"arena":"models","category":"3d","elo":1269,"win_rate":51.2,"rank":27},{"arena":"models","category":"asciiart","elo":1222,"win_rate":49.3,"rank":21},{"arena":"models","category":"codecategories","elo":1271,"win_rate":51.2,"rank":29},{"arena":"models","category":"dataviz","elo":1238,"win_rate":49.4,"rank":38},{"arena":"models","category":"gamedev","elo":1243,"win_rate":48.6,"rank":35},{"arena":"models","category":"svg","elo":1208,"win_rate":47.6,"rank":28},{"arena":"models","category":"uicomponent","elo":1282,"win_rate":52.6,"rank":25},{"arena":"models","category":"website","elo":1280,"win_rate":52.5,"rank":25}],"artificial_analysis":{"intelligence_index":26.3,"coding_index":60.8,"agentic_index":22.5}},"reasoning":{"mandatory":true,"default_enabled":true}},{"id":"~anthropic/claude-fable-latest","canonical_slug":"~anthropic/claude-fable-latest","alias_target":{"name":"Anthropic: Claude Fable 5.1","slug":"anthropic/claude-fable-5.1"},"hugging_face_id":null,"name":"Anthropic: Claude Fable Latest","created":1781029944,"description":"This model always redirects to the latest model in the Claude Fable family.","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.00001","completion":"0.00005","web_search":"0.01","input_cache_read":"0.00000025","input_cache_write":"0.0000125","input_cache_write_1h":"0.00002"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~anthropic/claude-fable-latest/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"anthropic/claude-fable-5","canonical_slug":"anthropic/claude-5-fable-20260609","hugging_face_id":null,"name":"Anthropic: Claude Fable 5","created":1781007515,"description":"Claude Fable 5 is a Mythos-class model from Anthropic, built for autonomous knowledge work and coding. It supports text, image, and file inputs with text output, with reasoning support and...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.00001","completion":"0.00005","web_search":"0.01","input_cache_read":"0.000001","input_cache_write":"0.0000125","input_cache_write_1h":"0.00002"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tool_choice","tools","verbosity"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-5-fable-20260609/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1285,"win_rate":61.9,"rank":1},{"arena":"agents","category":"agentichtmlslides","elo":1254,"win_rate":59.4,"rank":1},{"arena":"agents","category":"agenticslides(html)","elo":1252,"win_rate":59.5,"rank":1},{"arena":"agents","category":"androidnative","elo":1294,"win_rate":64.9,"rank":1},{"arena":"agents","category":"fullstack","elo":1282,"win_rate":59.5,"rank":6},{"arena":"agents","category":"godotgamedev","elo":1343,"win_rate":70,"rank":1},{"arena":"agents","category":"htmlslides","elo":1280,"win_rate":63.2,"rank":1},{"arena":"agents","category":"mobileapps","elo":1295,"win_rate":62.4,"rank":2},{"arena":"agents","category":"python-pptxslides","elo":1294,"win_rate":61.2,"rank":3},{"arena":"agents","category":"webapps","elo":1269,"win_rate":57.3,"rank":5},{"arena":"models","category":"3d","elo":1347,"win_rate":61.9,"rank":8},{"arena":"models","category":"asciiart","elo":1340,"win_rate":67.6,"rank":2},{"arena":"models","category":"codecategories","elo":1326,"win_rate":58.1,"rank":7},{"arena":"models","category":"dataviz","elo":1326,"win_rate":58.1,"rank":7},{"arena":"models","category":"gamedev","elo":1362,"win_rate":61.7,"rank":6},{"arena":"models","category":"svg","elo":1332,"win_rate":64.5,"rank":4},{"arena":"models","category":"uicomponent","elo":1331,"win_rate":58.1,"rank":9},{"arena":"models","category":"website","elo":1309,"win_rate":56.7,"rank":10}],"artificial_analysis":{"intelligence_index":49.7,"coding_index":76.5,"agentic_index":51}},"reasoning":{"mandatory":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"anthropic/claude-fable-5:batch","canonical_slug":"anthropic/claude-5-fable-20260609","hugging_face_id":null,"name":"Anthropic: Claude Fable 5 (batch)","created":1781007515,"description":"Claude Fable 5 is a Mythos-class model from Anthropic, built for autonomous knowledge work and coding. It supports text, image, and file inputs with text output, with reasoning support and...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000025","web_search":"0.01","input_cache_read":"0.0000005","input_cache_write":"0.00000625","input_cache_write_1h":"0.00001"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tool_choice","tools","verbosity"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-5-fable-20260609/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1285,"win_rate":61.9,"rank":1},{"arena":"agents","category":"agentichtmlslides","elo":1254,"win_rate":59.4,"rank":1},{"arena":"agents","category":"agenticslides(html)","elo":1252,"win_rate":59.5,"rank":1},{"arena":"agents","category":"androidnative","elo":1294,"win_rate":64.9,"rank":1},{"arena":"agents","category":"fullstack","elo":1282,"win_rate":59.5,"rank":6},{"arena":"agents","category":"godotgamedev","elo":1343,"win_rate":70,"rank":1},{"arena":"agents","category":"htmlslides","elo":1280,"win_rate":63.2,"rank":1},{"arena":"agents","category":"mobileapps","elo":1295,"win_rate":62.4,"rank":2},{"arena":"agents","category":"python-pptxslides","elo":1294,"win_rate":61.2,"rank":3},{"arena":"agents","category":"webapps","elo":1269,"win_rate":57.3,"rank":5},{"arena":"models","category":"3d","elo":1347,"win_rate":61.9,"rank":8},{"arena":"models","category":"asciiart","elo":1340,"win_rate":67.6,"rank":2},{"arena":"models","category":"codecategories","elo":1326,"win_rate":58.1,"rank":7},{"arena":"models","category":"dataviz","elo":1326,"win_rate":58.1,"rank":7},{"arena":"models","category":"gamedev","elo":1362,"win_rate":61.7,"rank":6},{"arena":"models","category":"svg","elo":1332,"win_rate":64.5,"rank":4},{"arena":"models","category":"uicomponent","elo":1331,"win_rate":58.1,"rank":9},{"arena":"models","category":"website","elo":1309,"win_rate":56.7,"rank":10}],"artificial_analysis":{"intelligence_index":49.7,"coding_index":76.5,"agentic_index":51}},"reasoning":{"mandatory":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"nvidia/nemotron-3.5-content-safety","canonical_slug":"nvidia/nemotron-3.5-content-safety-20260604","hugging_face_id":"nvidia/Nemotron-3.5-Content-Safety","name":"NVIDIA: Nemotron 3.5 Content Safety","created":1780581864,"description":"NVIDIA Nemotron 3.5 Content Safety is a compact 4B-parameter multimodal guardrail model from NVIDIA, fine-tuned from Google Gemma-3-4B. It moderates both inputs to and responses from LLMs and VLMs, accepting...","context_length":131072,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.0000002"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","temperature","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nvidia/nemotron-3.5-content-safety-20260604/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"nvidia/nemotron-3.5-content-safety:free","canonical_slug":"nvidia/nemotron-3.5-content-safety-20260604","hugging_face_id":"nvidia/Nemotron-3.5-Content-Safety","name":"NVIDIA: Nemotron 3.5 Content Safety (free)","created":1780581864,"description":"NVIDIA Nemotron 3.5 Content Safety is a compact 4B-parameter multimodal guardrail model from NVIDIA, fine-tuned from Google Gemma-3-4B. It moderates both inputs to and responses from LLMs and VLMs, accepting...","context_length":128000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":128000,"max_completion_tokens":8192,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","seed","temperature","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nvidia/nemotron-3.5-content-safety-20260604/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"nvidia/nemotron-3-ultra-550b-a55b","canonical_slug":"nvidia/nemotron-3-ultra-550b-a55b-20260604","hugging_face_id":"nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16","name":"NVIDIA: Nemotron 3 Ultra","created":1780551208,"description":"NVIDIA Nemotron 3 Ultra is an open frontier-reasoning and orchestration model from NVIDIA, with 55B active parameters out of 550B total (MoE). Built on a hybrid Transformer-Mamba mixture-of-experts architecture, it...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000006","completion":"0.0000024","input_cache_read":"0.00000012"},"top_provider":{"context_length":202800,"max_completion_tokens":182520,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nvidia/nemotron-3-ultra-550b-a55b-20260604/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1163,"win_rate":40.5,"rank":62},{"arena":"models","category":"asciiart","elo":1098,"win_rate":36.5,"rank":60},{"arena":"models","category":"codecategories","elo":1155,"win_rate":36.3,"rank":77},{"arena":"models","category":"dataviz","elo":1157,"win_rate":38.4,"rank":73},{"arena":"models","category":"gamedev","elo":1155,"win_rate":36.9,"rank":70},{"arena":"models","category":"svg","elo":1099,"win_rate":35.5,"rank":60},{"arena":"models","category":"uicomponent","elo":1151,"win_rate":37.2,"rank":73},{"arena":"models","category":"website","elo":1145,"win_rate":34.5,"rank":84}],"artificial_analysis":{"intelligence_index":23.4,"coding_index":49.3,"agentic_index":21.7}},"reasoning":{"mandatory":false,"default_enabled":true,"supports_max_tokens":true,"supported_efforts":["high","medium"],"default_effort":"high"}},{"id":"nvidia/nemotron-3-ultra-550b-a55b:free","canonical_slug":"nvidia/nemotron-3-ultra-550b-a55b-20260604","hugging_face_id":"nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16","name":"NVIDIA: Nemotron 3 Ultra (free)","created":1780551208,"description":"NVIDIA Nemotron 3 Ultra is an open frontier-reasoning and orchestration model from NVIDIA, with 55B active parameters out of 550B total (MoE). Built on a hybrid Transformer-Mamba mixture-of-experts architecture, it...","context_length":1000000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":1000000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","seed","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nvidia/nemotron-3-ultra-550b-a55b-20260604/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1163,"win_rate":40.5,"rank":62},{"arena":"models","category":"asciiart","elo":1098,"win_rate":36.5,"rank":60},{"arena":"models","category":"codecategories","elo":1155,"win_rate":36.3,"rank":77},{"arena":"models","category":"dataviz","elo":1157,"win_rate":38.4,"rank":73},{"arena":"models","category":"gamedev","elo":1155,"win_rate":36.9,"rank":70},{"arena":"models","category":"svg","elo":1099,"win_rate":35.5,"rank":60},{"arena":"models","category":"uicomponent","elo":1151,"win_rate":37.2,"rank":73},{"arena":"models","category":"website","elo":1145,"win_rate":34.5,"rank":84}],"artificial_analysis":{"intelligence_index":23.4,"coding_index":49.3,"agentic_index":21.7}},"reasoning":{"mandatory":false,"default_enabled":true,"supports_max_tokens":true,"supported_efforts":["high","medium"],"default_effort":"high"}},{"id":"qwen/qwen3.7-plus","canonical_slug":"qwen/qwen3.7-plus-20260602","hugging_face_id":null,"name":"Qwen: Qwen3.7 Plus","created":1780491783,"description":"Qwen3.7-Plus is a cost-effective model in Alibaba's Qwen3.7 series. It supports text and image input with text output, building on the series' text capabilities with a comprehensive upgrade to its...","context_length":1000000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.00000032","completion":"0.00000128","input_cache_read":"0.000000064","input_cache_write":"0.0000004","overrides":[{"min_prompt_tokens":256000,"prompt":"0.00000096","completion":"0.00000384","input_cache_read":"0.000000192","input_cache_write":"0.0000012"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.7-plus-20260602/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1267,"win_rate":48.9,"rank":28},{"arena":"models","category":"asciiart","elo":1147,"win_rate":40.3,"rank":46},{"arena":"models","category":"codecategories","elo":1281,"win_rate":50.6,"rank":24},{"arena":"models","category":"dataviz","elo":1293,"win_rate":52.7,"rank":14},{"arena":"models","category":"gamedev","elo":1287,"win_rate":50.4,"rank":21},{"arena":"models","category":"svg","elo":1241,"win_rate":50.2,"rank":18},{"arena":"models","category":"uicomponent","elo":1270,"win_rate":49.2,"rank":31},{"arena":"models","category":"website","elo":1281,"win_rate":51.2,"rank":24}],"artificial_analysis":{"intelligence_index":25.8,"coding_index":55.9,"agentic_index":19.7}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"minimax/minimax-m3","canonical_slug":"minimax/minimax-m3-20260531","hugging_face_id":"MiniMaxAI/Minimax-M3","name":"MiniMax: MiniMax M3","created":1780245374,"description":"MiniMax-M3 is a multimodal foundation model from MiniMax. It supports text, image, and video inputs with text output, a 1M-token context window, and is suited for long-horizon agentic work, coding,...","context_length":1048576,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000012","input_cache_read":"0.00000006"},"top_provider":{"context_length":524288,"max_completion_tokens":512000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/minimax/minimax-m3-20260531/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1159,"win_rate":44.7,"rank":19},{"arena":"agents","category":"androidnative","elo":1165,"win_rate":43.6,"rank":23},{"arena":"agents","category":"fullstack","elo":1207,"win_rate":48.8,"rank":19},{"arena":"agents","category":"htmlslides","elo":1182,"win_rate":46.5,"rank":11},{"arena":"agents","category":"mobileapps","elo":1190,"win_rate":46.2,"rank":20},{"arena":"agents","category":"python-pptxslides","elo":1207,"win_rate":47.1,"rank":14},{"arena":"agents","category":"webapps","elo":1214,"win_rate":48.5,"rank":19},{"arena":"models","category":"3d","elo":1239,"win_rate":51.5,"rank":35},{"arena":"models","category":"asciiart","elo":1181,"win_rate":47.6,"rank":31},{"arena":"models","category":"codecategories","elo":1265,"win_rate":52,"rank":31},{"arena":"models","category":"dataviz","elo":1250,"win_rate":51.6,"rank":33},{"arena":"models","category":"gamedev","elo":1237,"win_rate":47.1,"rank":37},{"arena":"models","category":"svg","elo":1194,"win_rate":48.2,"rank":32},{"arena":"models","category":"uicomponent","elo":1261,"win_rate":51.6,"rank":32},{"arena":"models","category":"website","elo":1271,"win_rate":52.6,"rank":29}],"artificial_analysis":{"intelligence_index":29.6,"coding_index":58.6,"agentic_index":30.8}},"reasoning":{"mandatory":false}},{"id":"minimax/minimax-m3:batch","canonical_slug":"minimax/minimax-m3-20260531","hugging_face_id":"MiniMaxAI/Minimax-M3","name":"MiniMax: MiniMax M3 (batch)","created":1780245374,"description":"MiniMax-M3 is a multimodal foundation model from MiniMax. It supports text, image, and video inputs with text output, a 1M-token context window, and is suited for long-horizon agentic work, coding,...","context_length":524288,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000012","input_cache_read":"0.00000006"},"top_provider":{"context_length":524288,"max_completion_tokens":471859,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/minimax/minimax-m3-20260531/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1159,"win_rate":44.7,"rank":19},{"arena":"agents","category":"androidnative","elo":1165,"win_rate":43.6,"rank":23},{"arena":"agents","category":"fullstack","elo":1207,"win_rate":48.8,"rank":19},{"arena":"agents","category":"htmlslides","elo":1182,"win_rate":46.5,"rank":11},{"arena":"agents","category":"mobileapps","elo":1190,"win_rate":46.2,"rank":20},{"arena":"agents","category":"python-pptxslides","elo":1207,"win_rate":47.1,"rank":14},{"arena":"agents","category":"webapps","elo":1214,"win_rate":48.5,"rank":19},{"arena":"models","category":"3d","elo":1239,"win_rate":51.5,"rank":35},{"arena":"models","category":"asciiart","elo":1181,"win_rate":47.6,"rank":31},{"arena":"models","category":"codecategories","elo":1265,"win_rate":52,"rank":31},{"arena":"models","category":"dataviz","elo":1250,"win_rate":51.6,"rank":33},{"arena":"models","category":"gamedev","elo":1237,"win_rate":47.1,"rank":37},{"arena":"models","category":"svg","elo":1194,"win_rate":48.2,"rank":32},{"arena":"models","category":"uicomponent","elo":1261,"win_rate":51.6,"rank":32},{"arena":"models","category":"website","elo":1271,"win_rate":52.6,"rank":29}],"artificial_analysis":{"intelligence_index":29.6,"coding_index":58.6,"agentic_index":30.8}},"reasoning":{"mandatory":false}},{"id":"stepfun/step-3.7-flash","canonical_slug":"stepfun/step-3.7-flash-20260528","hugging_face_id":"stepfun-ai/Step-3.7-Flash","name":"StepFun: Step 3.7 Flash","created":1779985069,"description":"Step 3.7 Flash is StepFun's latest high-efficiency multimodal Mixture-of-Experts model. It pairs a 196B-parameter language backbone with a vision encoder for native image and video understanding, activating roughly 11B parameters...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.00000115","input_cache_read":"0.00000004"},"top_provider":{"context_length":256000,"max_completion_tokens":230400,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/stepfun/step-3.7-flash-20260528/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1156,"win_rate":41.4,"rank":65},{"arena":"models","category":"asciiart","elo":1165,"win_rate":44.6,"rank":38},{"arena":"models","category":"codecategories","elo":1195,"win_rate":43.5,"rank":56},{"arena":"models","category":"dataviz","elo":1186,"win_rate":43.7,"rank":60},{"arena":"models","category":"gamedev","elo":1177,"win_rate":39.4,"rank":60},{"arena":"models","category":"svg","elo":1098,"win_rate":37.4,"rank":61},{"arena":"models","category":"uicomponent","elo":1196,"win_rate":43.6,"rank":54},{"arena":"models","category":"website","elo":1206,"win_rate":44.9,"rank":55}],"artificial_analysis":{"intelligence_index":null,"coding_index":39.6,"agentic_index":null}},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"anthropic/claude-opus-4.8","canonical_slug":"anthropic/claude-4.8-opus-20260528","hugging_face_id":null,"name":"Anthropic: Claude Opus 4.8","created":1779905091,"description":"Claude Opus 4.8 is Anthropic's most capable generally available model in the Opus family. It supports text, image, and file inputs with text output, with reasoning support and a 1M-token...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000025","web_search":"0.01","input_cache_read":"0.0000005","input_cache_write":"0.00000625","input_cache_write_1h":"0.00001"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.8-opus-20260528/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1238,"win_rate":55.8,"rank":4},{"arena":"agents","category":"agentichtmlslides","elo":1227,"win_rate":55.6,"rank":3},{"arena":"agents","category":"agenticslides","elo":1294,"win_rate":64.8,"rank":1},{"arena":"agents","category":"agenticslides(html)","elo":1230,"win_rate":56,"rank":3},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1310,"win_rate":68.9,"rank":1},{"arena":"agents","category":"androidnative","elo":1248,"win_rate":56.6,"rank":7},{"arena":"agents","category":"fullstack","elo":1258,"win_rate":57.1,"rank":10},{"arena":"agents","category":"godotgamedev","elo":1255,"win_rate":59,"rank":4},{"arena":"agents","category":"htmlslides","elo":1223,"win_rate":58,"rank":4},{"arena":"agents","category":"mobileapps","elo":1250,"win_rate":57.1,"rank":8},{"arena":"agents","category":"pptxslides","elo":1306,"win_rate":67.9,"rank":1},{"arena":"agents","category":"python-pptxslides","elo":1298,"win_rate":65.9,"rank":2},{"arena":"agents","category":"webapps","elo":1243,"win_rate":52.7,"rank":10},{"arena":"models","category":"3d","elo":1251,"win_rate":52.9,"rank":32},{"arena":"models","category":"asciiart","elo":1292,"win_rate":62.3,"rank":8},{"arena":"models","category":"codecategories","elo":1266,"win_rate":53.3,"rank":30},{"arena":"models","category":"dataviz","elo":1254,"win_rate":53.7,"rank":30},{"arena":"models","category":"gamedev","elo":1279,"win_rate":53.7,"rank":26},{"arena":"models","category":"svg","elo":1212,"win_rate":52.6,"rank":26},{"arena":"models","category":"uicomponent","elo":1271,"win_rate":54.3,"rank":29},{"arena":"models","category":"website","elo":1267,"win_rate":53.5,"rank":31}],"artificial_analysis":{"intelligence_index":42,"coding_index":74.3,"agentic_index":42.6}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"anthropic/claude-opus-4.8:batch","canonical_slug":"anthropic/claude-4.8-opus-20260528","hugging_face_id":null,"name":"Anthropic: Claude Opus 4.8 (batch)","created":1779905091,"description":"Claude Opus 4.8 is Anthropic's most capable generally available model in the Opus family. It supports text, image, and file inputs with text output, with reasoning support and a 1M-token...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.0000125","web_search":"0.01","input_cache_read":"0.00000025","input_cache_write":"0.000003125","input_cache_write_1h":"0.000005"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.8-opus-20260528/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1238,"win_rate":55.8,"rank":4},{"arena":"agents","category":"agentichtmlslides","elo":1227,"win_rate":55.6,"rank":3},{"arena":"agents","category":"agenticslides","elo":1294,"win_rate":64.8,"rank":1},{"arena":"agents","category":"agenticslides(html)","elo":1230,"win_rate":56,"rank":3},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1310,"win_rate":68.9,"rank":1},{"arena":"agents","category":"androidnative","elo":1248,"win_rate":56.6,"rank":7},{"arena":"agents","category":"fullstack","elo":1258,"win_rate":57.1,"rank":10},{"arena":"agents","category":"godotgamedev","elo":1255,"win_rate":59,"rank":4},{"arena":"agents","category":"htmlslides","elo":1223,"win_rate":58,"rank":4},{"arena":"agents","category":"mobileapps","elo":1250,"win_rate":57.1,"rank":8},{"arena":"agents","category":"pptxslides","elo":1306,"win_rate":67.9,"rank":1},{"arena":"agents","category":"python-pptxslides","elo":1298,"win_rate":65.9,"rank":2},{"arena":"agents","category":"webapps","elo":1243,"win_rate":52.7,"rank":10},{"arena":"models","category":"3d","elo":1251,"win_rate":52.9,"rank":32},{"arena":"models","category":"asciiart","elo":1292,"win_rate":62.3,"rank":8},{"arena":"models","category":"codecategories","elo":1266,"win_rate":53.3,"rank":30},{"arena":"models","category":"dataviz","elo":1254,"win_rate":53.7,"rank":30},{"arena":"models","category":"gamedev","elo":1279,"win_rate":53.7,"rank":26},{"arena":"models","category":"svg","elo":1212,"win_rate":52.6,"rank":26},{"arena":"models","category":"uicomponent","elo":1271,"win_rate":54.3,"rank":29},{"arena":"models","category":"website","elo":1267,"win_rate":53.5,"rank":31}],"artificial_analysis":{"intelligence_index":42,"coding_index":74.3,"agentic_index":42.6}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"qwen/qwen3.7-max","canonical_slug":"qwen/qwen3.7-max-20260520","hugging_face_id":null,"name":"Qwen: Qwen3.7 Max","created":1779376861,"description":"Qwen3.7-Max is the flagship model in Alibaba's Qwen3.7 series. It supports text input and output and is designed for agent-centric workloads, with particular strengths in coding, office and productivity tasks,...","context_length":1000000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.000001475","completion":"0.000004425","input_cache_read":"0.000000295","input_cache_write":"0.00000184375"},"top_provider":{"context_length":1000000,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.7-max-20260520/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1161,"win_rate":46.6,"rank":18},{"arena":"agents","category":"androidnative","elo":1170,"win_rate":48.6,"rank":22},{"arena":"agents","category":"fullstack","elo":1201,"win_rate":49.8,"rank":20},{"arena":"agents","category":"godotgamedev","elo":1225,"win_rate":55.6,"rank":7},{"arena":"agents","category":"htmlslides","elo":1161,"win_rate":43.5,"rank":15},{"arena":"agents","category":"mobileapps","elo":1174,"win_rate":45.5,"rank":25},{"arena":"agents","category":"python-pptxslides","elo":1194,"win_rate":47.2,"rank":17},{"arena":"agents","category":"webapps","elo":1224,"win_rate":50.3,"rank":15},{"arena":"models","category":"3d","elo":1297,"win_rate":55.4,"rank":18},{"arena":"models","category":"asciiart","elo":1232,"win_rate":53.6,"rank":18},{"arena":"models","category":"codecategories","elo":1290,"win_rate":55,"rank":20},{"arena":"models","category":"dataviz","elo":1291,"win_rate":54.5,"rank":15},{"arena":"models","category":"gamedev","elo":1292,"win_rate":54.5,"rank":18},{"arena":"models","category":"svg","elo":1244,"win_rate":58,"rank":16},{"arena":"models","category":"uicomponent","elo":1285,"win_rate":54.5,"rank":23},{"arena":"models","category":"website","elo":1285,"win_rate":55.1,"rank":21}],"artificial_analysis":{"intelligence_index":29.9,"coding_index":66,"agentic_index":23.9}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"x-ai/grok-build-0.1","canonical_slug":"x-ai/grok-build-0.1-20260520","hugging_face_id":null,"name":"SpaceXAI: Grok Build 0.1","created":1779298123,"description":"Grok Build 0.1 is SpaceXAI’s fast coding model trained specifically for agentic software engineering workflows. It supports text and image inputs with text output, and is optimized for interactive coding...","context_length":256000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Grok","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000002","web_search":"0.005","input_cache_read":"0.0000002","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000002","completion":"0.000004","input_cache_read":"0.0000004"}]},"top_provider":{"context_length":256000,"max_completion_tokens":230400,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","logprobs","max_tokens","reasoning","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/x-ai/grok-build-0.1-20260520/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":51.5,"agentic_index":null}},"reasoning":{"mandatory":true}},{"id":"google/gemini-3.5-flash","canonical_slug":"google/gemini-3.5-flash-20260519","hugging_face_id":null,"name":"Google: Gemini 3.5 Flash","created":1779193800,"description":"Gemini 3.5 Flash is Google's high-efficiency multimodal model, bringing near-Pro level coding and reasoning at Flash-tier cost and speed. It is highly optimized for coding proficiency and parallel agentic execution...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.0000015","completion":"0.000009","image":"0.0000015","audio":"0.000003","input_audio_cache":"0.0000003","web_search":"0.014","internal_reasoning":"0.000009","input_cache_read":"0.00000015","input_cache_write":"0.0000000833333333333333"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-01","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.5-flash-20260519/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1180,"win_rate":55.4,"rank":12},{"arena":"agents","category":"agentichtmlslides","elo":1162,"win_rate":45.8,"rank":6},{"arena":"agents","category":"agenticslides","elo":1244,"win_rate":57.5,"rank":3},{"arena":"agents","category":"agenticslides(html)","elo":1162,"win_rate":45.7,"rank":6},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1242,"win_rate":57.8,"rank":2},{"arena":"agents","category":"androidnative","elo":1199,"win_rate":57.9,"rank":14},{"arena":"agents","category":"fullstack","elo":1207,"win_rate":57.1,"rank":18},{"arena":"agents","category":"godotgamedev","elo":1137,"win_rate":43.1,"rank":20},{"arena":"agents","category":"htmlslides","elo":1145,"win_rate":46.5,"rank":19},{"arena":"agents","category":"mobileapps","elo":1196,"win_rate":51.8,"rank":17},{"arena":"agents","category":"pptxslides","elo":1244,"win_rate":57.7,"rank":2},{"arena":"agents","category":"python-pptxslides","elo":1247,"win_rate":57.4,"rank":9},{"arena":"agents","category":"webapps","elo":1217,"win_rate":53.1,"rank":17},{"arena":"models","category":"3d","elo":1271,"win_rate":56.8,"rank":26},{"arena":"models","category":"asciiart","elo":1274,"win_rate":59.5,"rank":12},{"arena":"models","category":"codecategories","elo":1278,"win_rate":54.7,"rank":26},{"arena":"models","category":"dataviz","elo":1250,"win_rate":53.7,"rank":32},{"arena":"models","category":"gamedev","elo":1287,"win_rate":54.1,"rank":20},{"arena":"models","category":"svg","elo":1280,"win_rate":60.5,"rank":9},{"arena":"models","category":"uicomponent","elo":1287,"win_rate":55.4,"rank":22},{"arena":"models","category":"website","elo":1275,"win_rate":54,"rank":28}],"artificial_analysis":{"intelligence_index":33,"coding_index":70.1,"agentic_index":27.3}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"google/gemini-3.5-flash:batch","canonical_slug":"google/gemini-3.5-flash-20260519","hugging_face_id":null,"name":"Google: Gemini 3.5 Flash (batch)","created":1779193800,"description":"Gemini 3.5 Flash is Google's high-efficiency multimodal model, bringing near-Pro level coding and reasoning at Flash-tier cost and speed. It is highly optimized for coding proficiency and parallel agentic execution...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000075","completion":"0.0000045","image":"0.00000075","audio":"0.0000015","input_audio_cache":"0.00000015","web_search":"0.014","internal_reasoning":"0.0000045","input_cache_read":"0.000000075"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-01","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.5-flash-20260519/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1180,"win_rate":55.4,"rank":12},{"arena":"agents","category":"agentichtmlslides","elo":1162,"win_rate":45.8,"rank":6},{"arena":"agents","category":"agenticslides","elo":1244,"win_rate":57.5,"rank":3},{"arena":"agents","category":"agenticslides(html)","elo":1162,"win_rate":45.7,"rank":6},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1242,"win_rate":57.8,"rank":2},{"arena":"agents","category":"androidnative","elo":1199,"win_rate":57.9,"rank":14},{"arena":"agents","category":"fullstack","elo":1207,"win_rate":57.1,"rank":18},{"arena":"agents","category":"godotgamedev","elo":1137,"win_rate":43.1,"rank":20},{"arena":"agents","category":"htmlslides","elo":1145,"win_rate":46.5,"rank":19},{"arena":"agents","category":"mobileapps","elo":1196,"win_rate":51.8,"rank":17},{"arena":"agents","category":"pptxslides","elo":1244,"win_rate":57.7,"rank":2},{"arena":"agents","category":"python-pptxslides","elo":1247,"win_rate":57.4,"rank":9},{"arena":"agents","category":"webapps","elo":1217,"win_rate":53.1,"rank":17},{"arena":"models","category":"3d","elo":1271,"win_rate":56.8,"rank":26},{"arena":"models","category":"asciiart","elo":1274,"win_rate":59.5,"rank":12},{"arena":"models","category":"codecategories","elo":1278,"win_rate":54.7,"rank":26},{"arena":"models","category":"dataviz","elo":1250,"win_rate":53.7,"rank":32},{"arena":"models","category":"gamedev","elo":1287,"win_rate":54.1,"rank":20},{"arena":"models","category":"svg","elo":1280,"win_rate":60.5,"rank":9},{"arena":"models","category":"uicomponent","elo":1287,"win_rate":55.4,"rank":22},{"arena":"models","category":"website","elo":1275,"win_rate":54,"rank":28}],"artificial_analysis":{"intelligence_index":33,"coding_index":70.1,"agentic_index":27.3}},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"perceptron/perceptron-mk1","canonical_slug":"perceptron/perceptron-mk1-20260512","hugging_face_id":null,"name":"Perceptron: Perceptron Mk1","created":1778597029,"description":"Perceptron Mk1 (Mark One) is Perceptron's highest-quality vision-language model for video and embodied reasoning.** It accepts image and video inputs paired with natural language queries, and produces detailed visual understanding...","context_length":32768,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000015"},"top_provider":{"context_length":32768,"max_completion_tokens":8192,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","structured_outputs","temperature","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/perceptron/perceptron-mk1-20260512/endpoints"},"reasoning":{"mandatory":false}},{"id":"google/gemini-3.1-flash-lite","canonical_slug":"google/gemini-3.1-flash-lite-20260507","hugging_face_id":null,"name":"Google: Gemini 3.1 Flash Lite","created":1778168828,"description":"Gemini 3.1 Flash Lite is Google’s GA high-efficiency multimodal model optimized for low-latency, high-volume workloads. It supports text, image, video, audio, and PDF inputs, and is designed for lightweight agentic...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.0000015","image":"0.00000025","audio":"0.0000005","input_audio_cache":"0.00000005","web_search":"0.014","internal_reasoning":"0.0000015","input_cache_read":"0.000000025","input_cache_write":"0.0000000833333333333333"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.1-flash-lite-20260507/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"minimal"}},{"id":"google/gemini-3.1-flash-lite:batch","canonical_slug":"google/gemini-3.1-flash-lite-20260507","hugging_face_id":null,"name":"Google: Gemini 3.1 Flash Lite (batch)","created":1778168828,"description":"Gemini 3.1 Flash Lite is Google’s GA high-efficiency multimodal model optimized for low-latency, high-volume workloads. It supports text, image, video, audio, and PDF inputs, and is designed for lightweight agentic...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.000000125","completion":"0.00000075","image":"0.000000125","audio":"0.00000025","input_audio_cache":"0.000000025","web_search":"0.014","internal_reasoning":"0.00000075","input_cache_read":"0.0000000125"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.1-flash-lite-20260507/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"minimal"}},{"id":"openai/gpt-chat-latest","canonical_slug":"openai/gpt-chat-latest-20260505","hugging_face_id":null,"name":"OpenAI: GPT Chat Latest","created":1778000212,"description":"GPT Chat Latest points to OpenAI's stable API alias `chat-latest` that always resolves to the latest Instant chat model used in ChatGPT. As OpenAI rolls out new Instant model updates...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.00003","web_search":"0.01","input_cache_read":"0.0000005"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_tokens","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-chat-latest-20260505/endpoints"}},{"id":"x-ai/grok-4.3","canonical_slug":"x-ai/grok-4.3-20260430","hugging_face_id":null,"name":"SpaceXAI: Grok 4.3","created":1777591821,"description":"Grok 4.3 is a reasoning model from SpaceXAI. It accepts text and image inputs with text output, and is suited for agentic workflows, instruction-following tasks, and applications requiring high factual...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Grok","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.0000025","web_search":"0.005","input_cache_read":"0.0000002","overrides":[{"min_prompt_tokens":200000,"prompt":"0.0000025","completion":"0.000005","input_cache_read":"0.0000004"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":900000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","logprobs","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/x-ai/grok-4.3-20260430/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1007,"win_rate":28.7,"rank":24},{"arena":"agents","category":"agentichtmlslides","elo":1066,"win_rate":31.8,"rank":9},{"arena":"agents","category":"agenticslides","elo":1072,"win_rate":31.9,"rank":9},{"arena":"agents","category":"agenticslides(html)","elo":1066,"win_rate":31.7,"rank":9},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1068,"win_rate":32.4,"rank":9},{"arena":"agents","category":"androidnative","elo":974,"win_rate":22.6,"rank":36},{"arena":"agents","category":"fullstack","elo":1015,"win_rate":28.5,"rank":40},{"arena":"agents","category":"godotgamedev","elo":1035,"win_rate":31.2,"rank":28},{"arena":"agents","category":"htmlslides","elo":1011,"win_rate":28.4,"rank":23},{"arena":"agents","category":"mobileapps","elo":1090,"win_rate":35.3,"rank":40},{"arena":"agents","category":"pptxslides","elo":1071,"win_rate":32.5,"rank":8},{"arena":"agents","category":"python-pptxslides","elo":1072,"win_rate":30.8,"rank":26},{"arena":"agents","category":"webapps","elo":1142,"win_rate":44,"rank":27},{"arena":"models","category":"3d","elo":1153,"win_rate":41.8,"rank":66},{"arena":"models","category":"asciiart","elo":1157,"win_rate":45.2,"rank":43},{"arena":"models","category":"codecategories","elo":1201,"win_rate":44.9,"rank":52},{"arena":"models","category":"dataviz","elo":1198,"win_rate":46,"rank":55},{"arena":"models","category":"gamedev","elo":1197,"win_rate":45.3,"rank":51},{"arena":"models","category":"svg","elo":1110,"win_rate":39.5,"rank":58},{"arena":"models","category":"uicomponent","elo":1206,"win_rate":46,"rank":50},{"arena":"models","category":"website","elo":1206,"win_rate":45,"rank":54}],"artificial_analysis":{"intelligence_index":25.4,"coding_index":42.2,"agentic_index":17.2}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","medium","low","none"],"default_effort":"low"}},{"id":"x-ai/grok-4.3:batch","canonical_slug":"x-ai/grok-4.3-20260430","hugging_face_id":null,"name":"SpaceXAI: Grok 4.3 (batch)","created":1777591821,"description":"Grok 4.3 is a reasoning model from SpaceXAI. It accepts text and image inputs with text output, and is suited for agentic workflows, instruction-following tasks, and applications requiring high factual...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Grok","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000002","web_search":"0.005","input_cache_read":"0.00000016","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000002","completion":"0.000004","input_cache_read":"0.00000032"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":900000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","logprobs","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/x-ai/grok-4.3-20260430/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1007,"win_rate":28.7,"rank":24},{"arena":"agents","category":"agentichtmlslides","elo":1066,"win_rate":31.8,"rank":9},{"arena":"agents","category":"agenticslides","elo":1072,"win_rate":31.9,"rank":9},{"arena":"agents","category":"agenticslides(html)","elo":1066,"win_rate":31.7,"rank":9},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1068,"win_rate":32.4,"rank":9},{"arena":"agents","category":"androidnative","elo":974,"win_rate":22.6,"rank":36},{"arena":"agents","category":"fullstack","elo":1015,"win_rate":28.5,"rank":40},{"arena":"agents","category":"godotgamedev","elo":1035,"win_rate":31.2,"rank":28},{"arena":"agents","category":"htmlslides","elo":1011,"win_rate":28.4,"rank":23},{"arena":"agents","category":"mobileapps","elo":1090,"win_rate":35.3,"rank":40},{"arena":"agents","category":"pptxslides","elo":1071,"win_rate":32.5,"rank":8},{"arena":"agents","category":"python-pptxslides","elo":1072,"win_rate":30.8,"rank":26},{"arena":"agents","category":"webapps","elo":1142,"win_rate":44,"rank":27},{"arena":"models","category":"3d","elo":1153,"win_rate":41.8,"rank":66},{"arena":"models","category":"asciiart","elo":1157,"win_rate":45.2,"rank":43},{"arena":"models","category":"codecategories","elo":1201,"win_rate":44.9,"rank":52},{"arena":"models","category":"dataviz","elo":1198,"win_rate":46,"rank":55},{"arena":"models","category":"gamedev","elo":1197,"win_rate":45.3,"rank":51},{"arena":"models","category":"svg","elo":1110,"win_rate":39.5,"rank":58},{"arena":"models","category":"uicomponent","elo":1206,"win_rate":46,"rank":50},{"arena":"models","category":"website","elo":1206,"win_rate":45,"rank":54}],"artificial_analysis":{"intelligence_index":25.4,"coding_index":42.2,"agentic_index":17.2}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","medium","low","none"],"default_effort":"low"}},{"id":"mistralai/mistral-medium-3-5","canonical_slug":"mistralai/mistral-medium-3.5-20260430","hugging_face_id":null,"name":"Mistral: Mistral Medium 3.5","created":1777570439,"description":"Mistral Medium 3.5 is a dense 128B instruction-following model from Mistral AI. It supports text and image inputs with text output, and is designed for agentic workflows, coding, and complex...","context_length":262144,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.0000015","completion":"0.0000075"},"top_provider":{"context_length":262144,"max_completion_tokens":209715,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-medium-3.5-20260430/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":14.9,"coding_index":46.9,"agentic_index":9.4}},"reasoning":{"mandatory":false,"supported_efforts":["high","none"],"default_effort":"high"}},{"id":"mistralai/mistral-medium-3-5:batch","canonical_slug":"mistralai/mistral-medium-3.5-20260430","hugging_face_id":null,"name":"Mistral: Mistral Medium 3.5 (batch)","created":1777570439,"description":"Mistral Medium 3.5 is a dense 128B instruction-following model from Mistral AI. It supports text and image inputs with text output, and is designed for agentic workflows, coding, and complex...","context_length":262144,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.00000075","completion":"0.00000375"},"top_provider":{"context_length":262144,"max_completion_tokens":209715,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-medium-3.5-20260430/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":14.9,"coding_index":46.9,"agentic_index":9.4}},"reasoning":{"mandatory":false,"supported_efforts":["high","none"],"default_effort":"high"}},{"id":"nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free","canonical_slug":"nvidia/nemotron-3-nano-omni-30b-a3b-reasoning-20260428","hugging_face_id":"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16","name":"NVIDIA: Nemotron 3 Nano Omni (free)","created":1777393095,"description":"NVIDIA Nemotron™ 3 Nano Omni is a 30B-A3B open multimodal model designed to function as a perception and context sub-agent in enterprise agent systems. It accepts text, image, video, and...","context_length":256000,"architecture":{"modality":"text+image+audio+video->text","input_modalities":["text","audio","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":256000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","seed","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.6,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nvidia/nemotron-3-nano-omni-30b-a3b-reasoning-20260428/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":13.8,"agentic_index":null}},"reasoning":{"mandatory":false,"default_enabled":true,"supports_max_tokens":true}},{"id":"~anthropic/claude-haiku-latest","canonical_slug":"~anthropic/claude-haiku-latest","alias_target":{"name":"Anthropic: Claude Haiku 4.5","slug":"anthropic/claude-haiku-4.5"},"hugging_face_id":null,"name":"Anthropic: Claude Haiku Latest","created":1777318492,"description":"This model always redirects to the latest model in the Claude Haiku family.","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000005","web_search":"0.01","input_cache_read":"0.0000001","input_cache_write":"0.00000125","input_cache_write_1h":"0.000002"},"top_provider":{"context_length":200000,"max_completion_tokens":64000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~anthropic/claude-haiku-latest/endpoints"},"reasoning":{"mandatory":false}},{"id":"~openai/gpt-mini-latest","canonical_slug":"~openai/gpt-mini-latest","alias_target":{"name":"OpenAI: GPT-5.4 Mini","slug":"openai/gpt-5.4-mini"},"hugging_face_id":null,"name":"OpenAI: GPT Mini Latest","created":1777318471,"description":"This model always redirects to the latest model in the GPT Mini family.","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.00000075","completion":"0.0000045","web_search":"0.01","input_cache_read":"0.000000075"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-08-31","expiration_date":null,"links":{"details":"/api/v1/models/~openai/gpt-mini-latest/endpoints"},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"~google/gemini-pro-latest","canonical_slug":"~google/gemini-pro-latest","alias_target":{"name":"Google: Gemini 3.1 Pro Preview","slug":"google/gemini-3.1-pro-preview"},"hugging_face_id":null,"name":"Google: Gemini Pro Latest","created":1777318451,"description":"This model always redirects to the latest model in the Gemini Pro family.","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["audio","file","image","text","video"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000012","image":"0.000002","audio":"0.000002","input_audio_cache":"0.0000002","web_search":"0.014","internal_reasoning":"0.000012","input_cache_read":"0.0000002","input_cache_write":"0.000000375","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000004","completion":"0.000018","audio":"0.000004","input_audio_cache":"0.0000004","input_cache_read":"0.0000004"}]},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~google/gemini-pro-latest/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"~moonshotai/kimi-latest","canonical_slug":"~moonshotai/kimi-latest","alias_target":{"name":"MoonshotAI: Kimi K3","slug":"moonshotai/kimi-k3"},"hugging_face_id":null,"name":"MoonshotAI: Kimi Latest","created":1777318428,"description":"This model always redirects to the latest model in the Kimi family.","context_length":1048576,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.000001975","completion":"0.00001106","input_cache_read":"0.0000002291"},"top_provider":{"context_length":1048576,"max_completion_tokens":943718,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~moonshotai/kimi-latest/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","high","low"],"default_effort":"max"}},{"id":"~google/gemini-flash-latest","canonical_slug":"~google/gemini-flash-latest","alias_target":{"name":"Google: Gemini 3.8 Flash","slug":"google/gemini-3.8-flash"},"hugging_face_id":null,"name":"Google: Gemini Flash Latest","created":1777318398,"description":"This model always redirects to the latest model in the Gemini Flash family.","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.00000075","completion":"0.00000375","image":"0.00000075","audio":"0.00000075","input_audio_cache":"0.000000075","web_search":"0.014","internal_reasoning":"0.00000375","input_cache_read":"0.000000075","input_cache_write":"0.0000000416666666666667"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~google/gemini-flash-latest/endpoints"},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"~anthropic/claude-sonnet-latest","canonical_slug":"~anthropic/claude-sonnet-latest","alias_target":{"name":"Anthropic: Claude Sonnet 5","slug":"anthropic/claude-sonnet-5"},"hugging_face_id":null,"name":"Anthropic: Claude Sonnet Latest","created":1777318368,"description":"This model always redirects to the latest model in the Claude Sonnet family.","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.00001","web_search":"0.01","input_cache_read":"0.0000002","input_cache_write":"0.0000025","input_cache_write_1h":"0.000004"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~anthropic/claude-sonnet-latest/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"qwen/qwen3.5-plus-20260420","canonical_slug":"qwen/qwen3.5-plus-20260420","hugging_face_id":null,"name":"Qwen: Qwen3.5 Plus 2026-04-20","created":1777261368,"description":"Qwen3.5 Plus (April 2026) is a large-scale multimodal language model from Alibaba. It accepts text, image, and video input and produces text output, with a 1M token context window. This...","context_length":1000000,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000018","input_cache_write":"0.000000375","overrides":[{"min_prompt_tokens":256000,"prompt":"0.000000375","completion":"0.00000225","input_cache_write":"0.00000046875"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.5-plus-20260420/endpoints"},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3.6-flash","canonical_slug":"qwen/qwen3.6-flash","hugging_face_id":null,"name":"Qwen: Qwen3.6 Flash","created":1777261362,"description":"Qwen3.6 Flash is a fast, efficient language model from Alibaba's Qwen 3.6 series. It supports text, image, and video input with a 1M token context window. Tiered pricing kicks in...","context_length":1000000,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.0000001875","completion":"0.000001125","input_cache_write":"0.000000234375","overrides":[{"min_prompt_tokens":256000,"prompt":"0.00000075","completion":"0.000003","input_cache_write":"0.0000009375"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.6-flash/endpoints"},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3.6-35b-a3b","canonical_slug":"qwen/qwen3.6-35b-a3b-20260415","hugging_face_id":"Qwen/Qwen3.6-35B-A3B","name":"Qwen: Qwen3.6 35B A3B","created":1777260255,"description":"Qwen3.6-35B-A3B is an open-weight multimodal model from Alibaba Cloud with 35 billion total parameters and 3 billion active parameters per token. It uses a hybrid sparse mixture-of-experts architecture combining Gated...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000009","input_cache_read":"0.00000005"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":20},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.6-35b-a3b-20260415/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":18.8,"coding_index":41.9,"agentic_index":15}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"qwen/qwen3.6-max-preview","canonical_slug":"qwen/qwen3.6-max-preview-20260420","hugging_face_id":null,"name":"Qwen: Qwen3.6 Max Preview","created":1777260242,"description":"Qwen3.6-Max-Preview is a proprietary frontier model from Alibaba Cloud built on a sparse mixture-of-experts architecture with approximately 1 trillion total parameters. It is optimized for agentic coding, tool use, and...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.000001027","completion":"0.000006162","input_cache_write":"0.00000128375","overrides":[{"min_prompt_tokens":128000,"prompt":"0.00000158","completion":"0.00000948","input_cache_write":"0.000001975"}]},"top_provider":{"context_length":262144,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.6-max-preview-20260420/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"qwen/qwen3.6-27b","canonical_slug":"qwen/qwen3.6-27b-20260422","hugging_face_id":"Qwen/Qwen3.6-27B","name":"Qwen: Qwen3.6 27B","created":1777255064,"description":"Qwen3.6 27B is a dense 27-billion-parameter language model from the Qwen Team at Alibaba, released in April 2026. It features hybrid multimodal capabilities — accepting text, image, and video inputs...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.000002","input_cache_read":"0.00000003"},"top_provider":{"context_length":262144,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.6-27b-20260422/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":21.9,"coding_index":53.7,"agentic_index":20.1}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"openai/gpt-5.5-pro","canonical_slug":"openai/gpt-5.5-pro-20260423","hugging_face_id":"","name":"OpenAI: GPT-5.5 Pro","created":1777051896,"description":"GPT-5.5 Pro is OpenAI’s high-capability model optimized for deep reasoning and accuracy on complex, high-stakes workloads. It features a 1M+ token context window (922K input, 128K output) with support for...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00003","completion":"0.00018","web_search":"0.01","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00006","completion":"0.00027"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-12-01","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.5-pro-20260423/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium"],"default_effort":"medium"}},{"id":"openai/gpt-5.5-pro:batch","canonical_slug":"openai/gpt-5.5-pro-20260423","hugging_face_id":"","name":"OpenAI: GPT-5.5 Pro (batch)","created":1777051896,"description":"GPT-5.5 Pro is OpenAI’s high-capability model optimized for deep reasoning and accuracy on complex, high-stakes workloads. It features a 1M+ token context window (922K input, 128K output) with support for...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000015","completion":"0.00009","web_search":"0.01","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00003","completion":"0.000135"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-12-01","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.5-pro-20260423/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium"],"default_effort":"medium"}},{"id":"openai/gpt-5.5","canonical_slug":"openai/gpt-5.5-20260423","hugging_face_id":"","name":"OpenAI: GPT-5.5","created":1777051893,"description":"GPT-5.5 is OpenAI’s frontier model designed for complex professional workloads, building on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks. It features a 1M+ token...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.00003","web_search":"0.01","input_cache_read":"0.0000005","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00001","completion":"0.000045","input_cache_read":"0.000001"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-12-01","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.5-20260423/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1179,"win_rate":52.4,"rank":14},{"arena":"agents","category":"agentichtmlslides","elo":1084,"win_rate":34.2,"rank":8},{"arena":"agents","category":"agenticslides","elo":1150,"win_rate":43.5,"rank":6},{"arena":"agents","category":"agenticslides(html)","elo":1077,"win_rate":33.2,"rank":8},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1155,"win_rate":45.2,"rank":6},{"arena":"agents","category":"androidnative","elo":1176,"win_rate":50.9,"rank":21},{"arena":"agents","category":"fullstack","elo":1099,"win_rate":43,"rank":29},{"arena":"agents","category":"godotgamedev","elo":1210,"win_rate":52.2,"rank":9},{"arena":"agents","category":"htmlslides","elo":1070,"win_rate":35.6,"rank":22},{"arena":"agents","category":"mobileapps","elo":1169,"win_rate":50.4,"rank":27},{"arena":"agents","category":"pptxslides","elo":1157,"win_rate":45.3,"rank":6},{"arena":"agents","category":"python-pptxslides","elo":1152,"win_rate":43.3,"rank":23},{"arena":"agents","category":"webapps","elo":1130,"win_rate":42.6,"rank":30},{"arena":"models","category":"3d","elo":1227,"win_rate":50.9,"rank":40},{"arena":"models","category":"asciiart","elo":1276,"win_rate":60.2,"rank":11},{"arena":"models","category":"codecategories","elo":1274,"win_rate":54.2,"rank":28},{"arena":"models","category":"dataviz","elo":1274,"win_rate":56,"rank":22},{"arena":"models","category":"gamedev","elo":1320,"win_rate":58.9,"rank":12},{"arena":"models","category":"svg","elo":1262,"win_rate":57.5,"rank":12},{"arena":"models","category":"uicomponent","elo":1271,"win_rate":54.6,"rank":30},{"arena":"models","category":"website","elo":1269,"win_rate":53.3,"rank":30}],"artificial_analysis":{"intelligence_index":38.6,"coding_index":74.9,"agentic_index":37.3}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.5:batch","canonical_slug":"openai/gpt-5.5-20260423","hugging_face_id":"","name":"OpenAI: GPT-5.5 (batch)","created":1777051893,"description":"GPT-5.5 is OpenAI’s frontier model designed for complex professional workloads, building on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks. It features a 1M+ token...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.000015","web_search":"0.01","input_cache_read":"0.00000025","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000005","completion":"0.0000225","input_cache_read":"0.0000005"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-12-01","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.5-20260423/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1179,"win_rate":52.4,"rank":14},{"arena":"agents","category":"agentichtmlslides","elo":1084,"win_rate":34.2,"rank":8},{"arena":"agents","category":"agenticslides","elo":1150,"win_rate":43.5,"rank":6},{"arena":"agents","category":"agenticslides(html)","elo":1077,"win_rate":33.2,"rank":8},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1155,"win_rate":45.2,"rank":6},{"arena":"agents","category":"androidnative","elo":1176,"win_rate":50.9,"rank":21},{"arena":"agents","category":"fullstack","elo":1099,"win_rate":43,"rank":29},{"arena":"agents","category":"godotgamedev","elo":1210,"win_rate":52.2,"rank":9},{"arena":"agents","category":"htmlslides","elo":1070,"win_rate":35.6,"rank":22},{"arena":"agents","category":"mobileapps","elo":1169,"win_rate":50.4,"rank":27},{"arena":"agents","category":"pptxslides","elo":1157,"win_rate":45.3,"rank":6},{"arena":"agents","category":"python-pptxslides","elo":1152,"win_rate":43.3,"rank":23},{"arena":"agents","category":"webapps","elo":1130,"win_rate":42.6,"rank":30},{"arena":"models","category":"3d","elo":1227,"win_rate":50.9,"rank":40},{"arena":"models","category":"asciiart","elo":1276,"win_rate":60.2,"rank":11},{"arena":"models","category":"codecategories","elo":1274,"win_rate":54.2,"rank":28},{"arena":"models","category":"dataviz","elo":1274,"win_rate":56,"rank":22},{"arena":"models","category":"gamedev","elo":1320,"win_rate":58.9,"rank":12},{"arena":"models","category":"svg","elo":1262,"win_rate":57.5,"rank":12},{"arena":"models","category":"uicomponent","elo":1271,"win_rate":54.6,"rank":30},{"arena":"models","category":"website","elo":1269,"win_rate":53.3,"rank":30}],"artificial_analysis":{"intelligence_index":38.6,"coding_index":74.9,"agentic_index":37.3}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"deepseek/deepseek-v4-pro","canonical_slug":"deepseek/deepseek-v4-pro-20260423","hugging_face_id":"deepseek-ai/DeepSeek-V4-Pro","name":"DeepSeek: DeepSeek V4 Pro 0423","created":1777000679,"description":"DeepSeek V4 Pro is a large-scale Mixture-of-Experts model from DeepSeek with 1.6T total parameters and 49B activated parameters, supporting a 1M-token context window. It is designed for advanced reasoning, coding,...","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.0000016","completion":"0.0000032","input_cache_read":"0.000000135"},"top_provider":{"context_length":1048576,"max_completion_tokens":393216,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_completion_tokens","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":1},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v4-pro-20260423/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"fullstack","elo":948,"win_rate":22.1,"rank":43},{"arena":"agents","category":"godotgamedev","elo":1059,"win_rate":34,"rank":26},{"arena":"agents","category":"webapps","elo":1000,"win_rate":26.4,"rank":39},{"arena":"models","category":"3d","elo":1283,"win_rate":56.8,"rank":22},{"arena":"models","category":"asciiart","elo":1167,"win_rate":46.4,"rank":36},{"arena":"models","category":"codecategories","elo":1255,"win_rate":51.9,"rank":35},{"arena":"models","category":"dataviz","elo":1219,"win_rate":48.7,"rank":47},{"arena":"models","category":"gamedev","elo":1252,"win_rate":52.3,"rank":33},{"arena":"models","category":"svg","elo":1166,"win_rate":45.4,"rank":43},{"arena":"models","category":"uicomponent","elo":1238,"win_rate":50.6,"rank":40},{"arena":"models","category":"website","elo":1247,"win_rate":50.6,"rank":39}],"artificial_analysis":{"intelligence_index":30.9,"coding_index":59.4,"agentic_index":27.7}},"reasoning":{"mandatory":false,"supported_efforts":["xhigh","high"],"default_effort":"high"}},{"id":"deepseek/deepseek-v4-flash","canonical_slug":"deepseek/deepseek-v4-flash-20260423","hugging_face_id":"deepseek-ai/DeepSeek-V4-Flash","name":"DeepSeek: DeepSeek V4 Flash 0423","created":1777000666,"description":"DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model from DeepSeek with 284B total parameters and 13B activated parameters, supporting a 1M-token context window. It is designed for fast inference and...","context_length":1048576,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.00000007938","completion":"0.00000015876","input_cache_read":"0.000000015876"},"top_provider":{"context_length":1024000,"max_completion_tokens":384000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_completion_tokens","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_a","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v4-flash-20260423/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1217,"win_rate":49.3,"rank":44},{"arena":"models","category":"asciiart","elo":1129,"win_rate":42.8,"rank":53},{"arena":"models","category":"codecategories","elo":1220,"win_rate":48.9,"rank":47},{"arena":"models","category":"dataviz","elo":1143,"win_rate":40.7,"rank":79},{"arena":"models","category":"gamedev","elo":1220,"win_rate":50.2,"rank":43},{"arena":"models","category":"svg","elo":1181,"win_rate":48.4,"rank":36},{"arena":"models","category":"uicomponent","elo":1179,"win_rate":44.7,"rank":62},{"arena":"models","category":"website","elo":1220,"win_rate":49.1,"rank":48}],"artificial_analysis":{"intelligence_index":24.8,"coding_index":52,"agentic_index":27.9}},"reasoning":{"mandatory":false,"supported_efforts":["xhigh","high"],"default_effort":"high"}},{"id":"tencent/hy3-preview","canonical_slug":"tencent/hy3-preview-20260421","hugging_face_id":"tencent/Hy3-preview","name":"Tencent: Hy3 preview","created":1776878150,"description":"Hy3 preview is a high-efficiency Mixture-of-Experts model from Tencent designed for agentic workflows and production use. It supports configurable reasoning levels across disabled, low, and high modes, allowing it to...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000018","completion":"0.0000006","input_cache_read":"0.00000006"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","seed","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.9,"top_p":1,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/tencent/hy3-preview-20260421/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":25.8,"coding_index":58.8,"agentic_index":25.6}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","low","none"],"default_effort":"high"}},{"id":"xiaomi/mimo-v2.5-pro","canonical_slug":"xiaomi/mimo-v2.5-pro-20260422","hugging_face_id":"XiaomiMiMo/MiMo-V2.5-Pro","name":"Xiaomi: MiMo-V2.5-Pro","created":1776874273,"description":"MiMo-V2.5-Pro is Xiaomi’s flagship model, delivering strong performance in general agentic capabilities, complex software engineering, and long-horizon tasks, with top rankings on benchmarks such as ClawEval, GDPVal, and SWE-bench Pro....","context_length":1050000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000435","completion":"0.00000087","input_cache_read":"0.0000000036"},"top_provider":{"context_length":1048576,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/xiaomi/mimo-v2.5-pro-20260422/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1277,"win_rate":55.4,"rank":25},{"arena":"models","category":"asciiart","elo":1169,"win_rate":46.7,"rank":35},{"arena":"models","category":"codecategories","elo":1287,"win_rate":53.2,"rank":22},{"arena":"models","category":"dataviz","elo":1280,"win_rate":52,"rank":19},{"arena":"models","category":"gamedev","elo":1285,"win_rate":54.7,"rank":23},{"arena":"models","category":"svg","elo":1212,"win_rate":50.6,"rank":27},{"arena":"models","category":"uicomponent","elo":1275,"win_rate":54.3,"rank":27},{"arena":"models","category":"website","elo":1285,"win_rate":52.3,"rank":19}],"artificial_analysis":{"intelligence_index":26.4,"coding_index":60.2,"agentic_index":22.7}},"reasoning":{"mandatory":false}},{"id":"xiaomi/mimo-v2.5","canonical_slug":"xiaomi/mimo-v2.5-20260422","hugging_face_id":"XiaomiMiMo/MiMo-V2.5","name":"Xiaomi: MiMo-V2.5","created":1776874269,"description":"MiMo-V2.5 is a native omnimodal model by Xiaomi. It delivers Pro-level agentic performance at roughly half the inference cost, while surpassing MiMo-V2-Omni in multimodal perception across image and video understanding...","context_length":1050000,"architecture":{"modality":"text+image+audio+video->text","input_modalities":["text","audio","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000014","completion":"0.00000028","input_cache_read":"0.0000000028"},"top_provider":{"context_length":1048576,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/xiaomi/mimo-v2.5-20260422/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1241,"win_rate":51.1,"rank":34},{"arena":"models","category":"asciiart","elo":1162,"win_rate":46.3,"rank":39},{"arena":"models","category":"codecategories","elo":1275,"win_rate":54.2,"rank":27},{"arena":"models","category":"dataviz","elo":1267,"win_rate":54.6,"rank":23},{"arena":"models","category":"gamedev","elo":1269,"win_rate":54,"rank":29},{"arena":"models","category":"svg","elo":1201,"win_rate":51.3,"rank":31},{"arena":"models","category":"uicomponent","elo":1274,"win_rate":54.6,"rank":28},{"arena":"models","category":"website","elo":1279,"win_rate":54.3,"rank":26}],"artificial_analysis":{"intelligence_index":22.3,"coding_index":56.8,"agentic_index":17.4}},"reasoning":{"mandatory":false}},{"id":"openai/gpt-5.4-image-2","canonical_slug":"openai/gpt-5.4-image-2-20260421","hugging_face_id":"","name":"OpenAI: GPT-5.4 Image 2","created":1776797528,"description":"[GPT-5.4](https://openrouter.ai/openai/gpt-5.4) Image 2 combines OpenAI's GPT-5.4 model with state-of-the-art image generation capabilities from GPT Image 2. It enables rich multimodal workflows, allowing users to seamlessly move between reasoning, coding, and...","context_length":272000,"architecture":{"modality":"text+image+file->text+image","input_modalities":["image","text","file"],"output_modalities":["image","text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000008","completion":"0.000015","image_output":"0.00003","web_search":"0.01","input_cache_read":"0.000002"},"top_provider":{"context_length":272000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","presence_penalty","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","top_logprobs"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.4-image-2-20260421/endpoints"},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"~anthropic/claude-opus-latest","canonical_slug":"~anthropic/claude-opus-latest","alias_target":{"name":"Claude Opus 5","slug":"anthropic/claude-opus-5"},"hugging_face_id":"","name":"Anthropic: Claude Opus Latest","created":1776795361,"description":"This model always redirects to the latest model in the Claude Opus family.","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000025","web_search":"0.01","input_cache_read":"0.0000005","input_cache_write":"0.00000625","input_cache_write_1h":"0.00001"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/~anthropic/claude-opus-latest/endpoints"},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"openrouter/pareto-code","canonical_slug":"openrouter/pareto-code","hugging_face_id":"","name":"Pareto Code Router","created":1776747900,"description":"The Pareto Router maintains a tiered shortlist of strong coding models, ranked by [Artificial Analysis](https://artificialanalysis.ai/) coding percentiles. Set min_coding_score between 0 and 1 on the [pareto-router plugin](https://openrouter.ai/docs/guides/routing/routers/pareto-router#the-min_coding_score-parameter) to control how...","context_length":2000000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"-1","completion":"-1"},"top_provider":{"context_length":null,"max_completion_tokens":null,"is_moderated":false},"per_request_limits":null,"supported_parameters":[],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openrouter/pareto-code/endpoints"}},{"id":"moonshotai/kimi-k2.6","canonical_slug":"moonshotai/kimi-k2.6-20260420","hugging_face_id":"moonshotai/Kimi-K2.6","name":"MoonshotAI: Kimi K2.6","created":1776699402,"description":"Kimi K2.6 is Moonshot AI's next-generation multimodal model, designed for long-horizon coding, coding-driven UI/UX generation, and multi-agent orchestration. It handles complex end-to-end coding tasks across Python, Rust, and Go, and...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000095","completion":"0.000004","input_cache_read":"0.00000016"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","parallel_tool_calls","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/moonshotai/kimi-k2.6-20260420/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1141,"win_rate":47.5,"rank":20},{"arena":"agents","category":"agentichtmlslides","elo":1248,"win_rate":59,"rank":2},{"arena":"agents","category":"agenticslides","elo":1187,"win_rate":45.8,"rank":4},{"arena":"agents","category":"agenticslides(html)","elo":1252,"win_rate":59.2,"rank":2},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1186,"win_rate":45.5,"rank":4},{"arena":"agents","category":"androidnative","elo":1177,"win_rate":50.1,"rank":19},{"arena":"agents","category":"fullstack","elo":1164,"win_rate":50.9,"rank":25},{"arena":"agents","category":"godotgamedev","elo":1159,"win_rate":47,"rank":14},{"arena":"agents","category":"htmlslides","elo":1187,"win_rate":52.3,"rank":10},{"arena":"agents","category":"mobileapps","elo":1180,"win_rate":48.7,"rank":23},{"arena":"agents","category":"pptxslides","elo":1181,"win_rate":44.3,"rank":4},{"arena":"agents","category":"python-pptxslides","elo":1180,"win_rate":42.1,"rank":18},{"arena":"agents","category":"webapps","elo":1268,"win_rate":59.3,"rank":6},{"arena":"models","category":"3d","elo":1297,"win_rate":57.6,"rank":17},{"arena":"models","category":"asciiart","elo":1184,"win_rate":47.5,"rank":29},{"arena":"models","category":"codecategories","elo":1287,"win_rate":54.8,"rank":21},{"arena":"models","category":"dataviz","elo":1266,"win_rate":52,"rank":24},{"arena":"models","category":"gamedev","elo":1274,"win_rate":54.9,"rank":28},{"arena":"models","category":"svg","elo":1207,"win_rate":51.2,"rank":29},{"arena":"models","category":"uicomponent","elo":1282,"win_rate":55.6,"rank":24},{"arena":"models","category":"website","elo":1282,"win_rate":54.2,"rank":22}],"artificial_analysis":{"intelligence_index":null,"coding_index":61.8,"agentic_index":22.1}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"anthropic/claude-opus-4.7","canonical_slug":"anthropic/claude-4.7-opus-20260416","hugging_face_id":null,"name":"Anthropic: Claude Opus 4.7","created":1776351100,"description":"Opus 4.7 is the next generation of Anthropic's Opus family, built for long-running, asynchronous agents. Building on the coding and agentic strengths of Opus 4.6, it delivers stronger performance on...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000025","web_search":"0.01","input_cache_read":"0.0000005","input_cache_write":"0.00000625","input_cache_write_1h":"0.00001"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.7-opus-20260416/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":73.6,"agentic_index":39.5}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"anthropic/claude-opus-4.7:batch","canonical_slug":"anthropic/claude-4.7-opus-20260416","hugging_face_id":null,"name":"Anthropic: Claude Opus 4.7 (batch)","created":1776351100,"description":"Opus 4.7 is the next generation of Anthropic's Opus family, built for long-running, asynchronous agents. Building on the coding and agentic strengths of Opus 4.6, it delivers stronger performance on...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.0000125","web_search":"0.01","input_cache_read":"0.00000025","input_cache_write":"0.000003125","input_cache_write_1h":"0.000005"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.7-opus-20260416/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":73.6,"agentic_index":39.5}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["max","xhigh","high","medium","low"],"default_effort":"high"}},{"id":"z-ai/glm-5.1","canonical_slug":"z-ai/glm-5.1-20260406","hugging_face_id":"zai-org/GLM-5.1","name":"Z.ai: GLM 5.1","created":1775578025,"description":"GLM-5.1 delivers a major leap in coding capability, with particularly significant gains in handling long-horizon tasks. Unlike previous models built around minute-level interactions, GLM-5.1 can work independently and continuously on...","context_length":204800,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000966","completion":"0.000003036","input_cache_read":"0.0000001794"},"top_provider":{"context_length":200000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-5.1-20260406/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1336,"win_rate":62.6,"rank":10},{"arena":"models","category":"asciiart","elo":1146,"win_rate":43.8,"rank":47},{"arena":"models","category":"codecategories","elo":1282,"win_rate":53.8,"rank":23},{"arena":"models","category":"dataviz","elo":1366,"win_rate":67,"rank":1},{"arena":"models","category":"gamedev","elo":1285,"win_rate":56.6,"rank":22},{"arena":"models","category":"svg","elo":1242,"win_rate":57,"rank":17},{"arena":"models","category":"uicomponent","elo":1292,"win_rate":54.9,"rank":20},{"arena":"models","category":"website","elo":1290,"win_rate":55.1,"rank":17},{"arena":"agents","category":"agenticgamedev","elo":1165,"win_rate":49.8,"rank":17},{"arena":"agents","category":"agentichtmlslides","elo":1205,"win_rate":52,"rank":5},{"arena":"agents","category":"agenticslides","elo":1245,"win_rate":54.4,"rank":2},{"arena":"agents","category":"agenticslides(html)","elo":1204,"win_rate":51.8,"rank":5},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1240,"win_rate":53.3,"rank":3},{"arena":"agents","category":"androidnative","elo":1192,"win_rate":52,"rank":16},{"arena":"agents","category":"fullstack","elo":1175,"win_rate":53.6,"rank":23},{"arena":"agents","category":"godotgamedev","elo":1104,"win_rate":39.5,"rank":24},{"arena":"agents","category":"htmlslides","elo":1173,"win_rate":48.9,"rank":14},{"arena":"agents","category":"mobileapps","elo":1191,"win_rate":52,"rank":18},{"arena":"agents","category":"pptxslides","elo":1241,"win_rate":53.5,"rank":3},{"arena":"agents","category":"python-pptxslides","elo":1258,"win_rate":54.2,"rank":6},{"arena":"agents","category":"webapps","elo":1188,"win_rate":51.2,"rank":23}],"artificial_analysis":{"intelligence_index":26.4,"coding_index":55.8,"agentic_index":25.2}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"google/gemma-4-26b-a4b-it","canonical_slug":"google/gemma-4-26b-a4b-it-20260403","hugging_face_id":"google/gemma-4-26B-A4B-it","name":"Google: Gemma 4 26B A4B ","created":1775227989,"description":"Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["image","text","video"],"output_modalities":["text"],"tokenizer":"Gemma","instruct_type":null},"pricing":{"prompt":"0.00000009","completion":"0.0000003","input_cache_read":"0.00000005"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":64},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemma-4-26b-a4b-it-20260403/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":39.3,"agentic_index":null}},"reasoning":{"mandatory":false,"default_enabled":false}},{"id":"google/gemma-4-26b-a4b-it:free","canonical_slug":"google/gemma-4-26b-a4b-it-20260403","hugging_face_id":"google/gemma-4-26B-A4B-it","name":"Google: Gemma 4 26B A4B (free)","created":1775227989,"description":"Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["image","text","video"],"output_modalities":["text"],"tokenizer":"Gemma","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":262144,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":64},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemma-4-26b-a4b-it-20260403/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":39.3,"agentic_index":null}},"reasoning":{"mandatory":false,"default_enabled":false}},{"id":"google/gemma-4-31b-it","canonical_slug":"google/gemma-4-31b-it-20260402","hugging_face_id":"google/gemma-4-31B-it","name":"Google: Gemma 4 31B","created":1775148486,"description":"Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K token context window, configurable thinking/reasoning mode, native function...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["image","text","video"],"output_modalities":["text"],"tokenizer":"Gemma","instruct_type":null},"pricing":{"prompt":"0.00000009","completion":"0.00000034","input_cache_read":"0.00000005"},"top_provider":{"context_length":262144,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":64,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemma-4-31b-it-20260402/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":15.4,"coding_index":43.4,"agentic_index":6.7}},"reasoning":{"mandatory":false,"default_enabled":false}},{"id":"google/gemma-4-31b-it:batch","canonical_slug":"google/gemma-4-31b-it-20260402","hugging_face_id":"google/gemma-4-31B-it","name":"Google: Gemma 4 31B (batch)","created":1775148486,"description":"Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K token context window, configurable thinking/reasoning mode, native function...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["image","text","video"],"output_modalities":["text"],"tokenizer":"Gemma","instruct_type":null},"pricing":{"prompt":"0.00000039","completion":"0.00000097"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":64,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemma-4-31b-it-20260402/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":15.4,"coding_index":43.4,"agentic_index":6.7}},"reasoning":{"mandatory":false,"default_enabled":false}},{"id":"google/gemma-4-31b-it:free","canonical_slug":"google/gemma-4-31b-it-20260402","hugging_face_id":"google/gemma-4-31B-it","name":"Google: Gemma 4 31B (free)","created":1775148486,"description":"Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K token context window, configurable thinking/reasoning mode, native function...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["image","text","video"],"output_modalities":["text"],"tokenizer":"Gemma","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":262144,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":64,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemma-4-31b-it-20260402/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":15.4,"coding_index":43.4,"agentic_index":6.7}},"reasoning":{"mandatory":false,"default_enabled":false}},{"id":"qwen/qwen3.6-plus","canonical_slug":"qwen/qwen3.6-plus-04-02","hugging_face_id":"","name":"Qwen: Qwen3.6 Plus","created":1775133557,"description":"Qwen 3.6 Plus builds on a hybrid architecture that combines efficient linear attention with sparse mixture-of-experts routing, enabling strong scalability and high-performance inference. Compared to the 3.5 series, it delivers...","context_length":1000000,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.000000325","completion":"0.00000195","input_cache_write":"0.00000040625","overrides":[{"min_prompt_tokens":256000,"prompt":"0.0000013","completion":"0.0000039","input_cache_write":"0.000001625"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.6-plus-04-02/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1230,"win_rate":51.1,"rank":39},{"arena":"models","category":"asciiart","elo":1136,"win_rate":42.9,"rank":51},{"arena":"models","category":"codecategories","elo":1250,"win_rate":51.1,"rank":37},{"arena":"models","category":"dataviz","elo":1245,"win_rate":51.2,"rank":35},{"arena":"models","category":"gamedev","elo":1233,"win_rate":49.9,"rank":39},{"arena":"models","category":"svg","elo":1187,"win_rate":51.2,"rank":35},{"arena":"models","category":"uicomponent","elo":1249,"win_rate":51.7,"rank":38},{"arena":"models","category":"website","elo":1252,"win_rate":50.9,"rank":37}],"artificial_analysis":{"intelligence_index":null,"coding_index":54.5,"agentic_index":null}},"reasoning":{"mandatory":false}},{"id":"z-ai/glm-5v-turbo","canonical_slug":"z-ai/glm-5v-turbo-20260401","hugging_face_id":"","name":"Z.ai: GLM 5V Turbo","created":1775061458,"description":"GLM-5V-Turbo is Z.ai’s first native multimodal agent foundation model, built for vision-based coding and agent-driven tasks. It natively handles image, video, and text inputs, excels at long-horizon planning, complex coding,...","context_length":202752,"architecture":{"modality":"text+image+video->text","input_modalities":["image","text","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000012","completion":"0.000004","input_cache_read":"0.00000024"},"top_provider":{"context_length":202752,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":"2098-12-31","links":{"details":"/api/v1/models/z-ai/glm-5v-turbo-20260401/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1102,"win_rate":42.6,"rank":23},{"arena":"agents","category":"agentichtmlslides","elo":1134,"win_rate":41.7,"rank":7},{"arena":"agents","category":"agenticslides","elo":1171,"win_rate":51.6,"rank":5},{"arena":"agents","category":"agenticslides(html)","elo":1137,"win_rate":41.8,"rank":7},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1183,"win_rate":53.9,"rank":5},{"arena":"agents","category":"androidnative","elo":1267,"win_rate":54.8,"rank":3},{"arena":"agents","category":"fullstack","elo":1158,"win_rate":52,"rank":26},{"arena":"agents","category":"godotgamedev","elo":997,"win_rate":26.9,"rank":29},{"arena":"agents","category":"htmlslides","elo":1123,"win_rate":44.6,"rank":21},{"arena":"agents","category":"mobileapps","elo":1161,"win_rate":50.4,"rank":30},{"arena":"agents","category":"pptxslides","elo":1164,"win_rate":52.1,"rank":5},{"arena":"agents","category":"python-pptxslides","elo":1165,"win_rate":51.9,"rank":20},{"arena":"agents","category":"webapps","elo":1138,"win_rate":44.7,"rank":29},{"arena":"models","category":"3d","elo":1238,"win_rate":53.3,"rank":36},{"arena":"models","category":"asciiart","elo":1130,"win_rate":43,"rank":52},{"arena":"models","category":"codecategories","elo":1244,"win_rate":50.6,"rank":41},{"arena":"models","category":"dataviz","elo":1216,"win_rate":48,"rank":49},{"arena":"models","category":"gamedev","elo":1241,"win_rate":51.7,"rank":36},{"arena":"models","category":"svg","elo":1175,"win_rate":49.8,"rank":38},{"arena":"models","category":"uicomponent","elo":1227,"win_rate":49.3,"rank":43},{"arena":"models","category":"website","elo":1243,"win_rate":49.6,"rank":40}]},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"arcee-ai/trinity-large-thinking","canonical_slug":"arcee-ai/trinity-large-thinking","hugging_face_id":"arcee-ai/Trinity-Large-Thinking","name":"Arcee AI: Trinity Large Thinking","created":1775058318,"description":"Trinity Large Thinking is a powerful open source reasoning model from the team at Arcee AI. It shows strong performance in PinchBench, agentic workloads, and reasoning tasks. Launch video: https://youtu.be/Gc82AXLa0Rg?si=4RLn6WBz33qT--B7...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.0000008","input_cache_read":"0.00000006"},"top_provider":{"context_length":262144,"max_completion_tokens":80000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":0.3,"top_p":0.8,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/arcee-ai/trinity-large-thinking/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1110,"win_rate":41.3,"rank":83},{"arena":"models","category":"asciiart","elo":1059,"win_rate":37.1,"rank":63},{"arena":"models","category":"codecategories","elo":1132,"win_rate":40.1,"rank":84},{"arena":"models","category":"dataviz","elo":1114,"win_rate":39.3,"rank":88},{"arena":"models","category":"gamedev","elo":1102,"win_rate":38.4,"rank":89},{"arena":"models","category":"svg","elo":1041,"win_rate":35.2,"rank":75},{"arena":"models","category":"uicomponent","elo":1060,"win_rate":32.6,"rank":94},{"arena":"models","category":"website","elo":1149,"win_rate":41.3,"rank":81}],"artificial_analysis":{"intelligence_index":10.9,"coding_index":25.8,"agentic_index":1.2}},"reasoning":{"mandatory":true}},{"id":"x-ai/grok-4.20-multi-agent","canonical_slug":"x-ai/grok-4.20-multi-agent-20260309","hugging_face_id":"","name":"SpaceXAI: Grok 4.20 Multi-Agent","created":1774979158,"description":"Grok 4.20 Multi-Agent is a variant of SpaceXAI’s Grok 4.20 designed for collaborative, agent-based workflows. Multiple agents operate in parallel to conduct deep research, coordinate tool use, and synthesize information...","context_length":2000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Grok","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.0000025","web_search":"0.005","input_cache_read":"0.0000002","overrides":[{"min_prompt_tokens":200000,"prompt":"0.0000025","completion":"0.000005","input_cache_read":"0.0000004"}]},"top_provider":{"context_length":2000000,"max_completion_tokens":1800000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","logprobs","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","temperature","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-09-01","expiration_date":null,"links":{"details":"/api/v1/models/x-ai/grok-4.20-multi-agent-20260309/endpoints"},"reasoning":{"mandatory":true,"default_enabled":true,"supported_efforts":["xhigh","high","medium","low"],"default_effort":"medium"}},{"id":"x-ai/grok-4.20","canonical_slug":"x-ai/grok-4.20-20260309","hugging_face_id":"","name":"SpaceXAI: Grok 4.20","created":1774979019,"description":"Grok 4.20 is a reasoning model from SpaceXAI with industry-leading speed and agentic tool calling capabilities. It combines the lowest hallucination rate on the market with strict prompt adherance, delivering...","context_length":2000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Grok","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.0000025","web_search":"0.005","input_cache_read":"0.0000002","overrides":[{"min_prompt_tokens":200000,"prompt":"0.0000025","completion":"0.000005","input_cache_read":"0.0000004"}]},"top_provider":{"context_length":2000000,"max_completion_tokens":1800000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","logprobs","max_tokens","reasoning","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-09-01","expiration_date":null,"links":{"details":"/api/v1/models/x-ai/grok-4.20-20260309/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"androidnative","elo":1060,"win_rate":35.1,"rank":34},{"arena":"agents","category":"fullstack","elo":1062,"win_rate":41,"rank":34},{"arena":"agents","category":"godotgamedev","elo":1089,"win_rate":38.5,"rank":25},{"arena":"agents","category":"htmlslides","elo":1180,"win_rate":45.7,"rank":12},{"arena":"agents","category":"mobileapps","elo":1116,"win_rate":45.4,"rank":38},{"arena":"agents","category":"webapps","elo":1157,"win_rate":49.7,"rank":25},{"arena":"models","category":"3d","elo":1221,"win_rate":52.1,"rank":42},{"arena":"models","category":"asciiart","elo":1193,"win_rate":48.2,"rank":25},{"arena":"models","category":"codecategories","elo":1236,"win_rate":51.8,"rank":42},{"arena":"models","category":"dataviz","elo":1234,"win_rate":52,"rank":43},{"arena":"models","category":"gamedev","elo":1216,"win_rate":50.1,"rank":45},{"arena":"models","category":"svg","elo":1189,"win_rate":52.4,"rank":34},{"arena":"models","category":"uicomponent","elo":1214,"win_rate":48.8,"rank":47},{"arena":"models","category":"website","elo":1242,"win_rate":51.9,"rank":41}]},"reasoning":{"mandatory":false,"default_enabled":false}},{"id":"google/lyria-3-pro-preview","canonical_slug":"google/lyria-3-pro-preview-20260330","hugging_face_id":null,"name":"Google: Lyria 3 Pro Preview","created":1774907286,"description":"Full-length songs are priced at $0.08 per song. Lyria 3 is Google's family of music generation models, available through the Gemini API. With Lyria 3, you can generate high-quality, 48kHz...","context_length":1048576,"architecture":{"modality":"text+image->text+audio","input_modalities":["text","image"],"output_modalities":["text","audio"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_tokens","response_format","seed","temperature","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/lyria-3-pro-preview-20260330/endpoints"}},{"id":"google/lyria-3-clip-preview","canonical_slug":"google/lyria-3-clip-preview-20260330","hugging_face_id":null,"name":"Google: Lyria 3 Clip Preview","created":1774907255,"description":"30 second duration clips are priced at $0.04 per clip. Lyria 3 is Google's family of music generation models, available through the Gemini API. With Lyria 3, you can generate...","context_length":1048576,"architecture":{"modality":"text+image->text+audio","input_modalities":["text","image"],"output_modalities":["text","audio"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_tokens","response_format","seed","temperature","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/lyria-3-clip-preview-20260330/endpoints"}},{"id":"kwaipilot/kat-coder-pro-v2","canonical_slug":"kwaipilot/kat-coder-pro-v2-20260327","hugging_face_id":"","name":"Kwaipilot: KAT-Coder-Pro V2","created":1774649310,"description":"KAT-Coder-Pro V2 is the latest high-performance model in KwaiKAT’s KAT-Coder series, designed for complex enterprise-grade software engineering and SaaS integration. It builds on the agentic coding strengths of earlier versions,...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000012","input_cache_read":"0.00000006"},"top_provider":{"context_length":262144,"max_completion_tokens":144000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/kwaipilot/kat-coder-pro-v2-20260327/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":59.5,"agentic_index":null}}},{"id":"rekaai/reka-edge","canonical_slug":"rekaai/reka-edge-2603","hugging_face_id":"RekaAI/reka-edge-2603","name":"Reka Edge","created":1774026965,"description":"Reka Edge is an extremely efficient 7B multimodal vision-language model that accepts image/video+text inputs and generates text outputs. This model is optimized specifically to deliver industry-leading performance in image understanding,...","context_length":16384,"architecture":{"modality":"text+image+video->text","input_modalities":["image","text","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000001"},"top_provider":{"context_length":16384,"max_completion_tokens":14745,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logprobs","max_tokens","presence_penalty","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/rekaai/reka-edge-2603/endpoints"},"reasoning":{"mandatory":false}},{"id":"minimax/minimax-m2.7","canonical_slug":"minimax/minimax-m2.7-20260318","hugging_face_id":"MiniMaxAI/MiniMax-M2.7","name":"MiniMax: MiniMax M2.7","created":1773836697,"description":"MiniMax-M2.7 is a next-generation large language model designed for autonomous, real-world productivity and continuous improvement. Built to actively participate in its own evolution, M2.7 integrates advanced agentic capabilities through multi-agent...","context_length":204800,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000012","input_cache_read":"0.00000006"},"top_provider":{"context_length":204800,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/minimax/minimax-m2.7-20260318/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1221,"win_rate":50.3,"rank":41},{"arena":"models","category":"asciiart","elo":1158,"win_rate":47.4,"rank":42},{"arena":"models","category":"codecategories","elo":1249,"win_rate":51.7,"rank":38},{"arena":"models","category":"dataviz","elo":1249,"win_rate":52.8,"rank":34},{"arena":"models","category":"gamedev","elo":1231,"win_rate":51,"rank":40},{"arena":"models","category":"svg","elo":1162,"win_rate":48.6,"rank":46},{"arena":"models","category":"uicomponent","elo":1225,"win_rate":49.1,"rank":44},{"arena":"models","category":"website","elo":1257,"win_rate":52.2,"rank":36}],"artificial_analysis":{"intelligence_index":23.2,"coding_index":52.6,"agentic_index":16.8}},"reasoning":{"mandatory":true}},{"id":"openai/gpt-5.4-nano","canonical_slug":"openai/gpt-5.4-nano-20260317","hugging_face_id":"","name":"OpenAI: GPT-5.4 Nano","created":1773748187,"description":"GPT-5.4 nano is the most lightweight and cost-efficient variant of the GPT-5.4 family, optimized for speed-critical and high-volume tasks. It supports text and image inputs and is designed for low-latency...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.00000125","web_search":"0.01","input_cache_read":"0.00000002"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-08-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.4-nano-20260317/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":21.2,"coding_index":56.1,"agentic_index":17.7}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.4-nano:batch","canonical_slug":"openai/gpt-5.4-nano-20260317","hugging_face_id":"","name":"OpenAI: GPT-5.4 Nano (batch)","created":1773748187,"description":"GPT-5.4 nano is the most lightweight and cost-efficient variant of the GPT-5.4 family, optimized for speed-critical and high-volume tasks. It supports text and image inputs and is designed for low-latency...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.000000625","web_search":"0.01","input_cache_read":"0.00000001"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-08-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.4-nano-20260317/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":21.2,"coding_index":56.1,"agentic_index":17.7}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.4-mini","canonical_slug":"openai/gpt-5.4-mini-20260317","hugging_face_id":"","name":"OpenAI: GPT-5.4 Mini","created":1773748178,"description":"GPT-5.4 mini brings the core capabilities of GPT-5.4 to a faster, more efficient model optimized for high-throughput workloads. It supports text and image inputs with strong performance across reasoning, coding,...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000075","completion":"0.0000045","web_search":"0.01","input_cache_read":"0.000000075"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-08-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.4-mini-20260317/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":24.6,"coding_index":56.1,"agentic_index":19.7}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.4-mini:batch","canonical_slug":"openai/gpt-5.4-mini-20260317","hugging_face_id":"","name":"OpenAI: GPT-5.4 Mini (batch)","created":1773748178,"description":"GPT-5.4 mini brings the core capabilities of GPT-5.4 to a faster, more efficient model optimized for high-throughput workloads. It supports text and image inputs with strong performance across reasoning, coding,...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000000375","completion":"0.00000225","web_search":"0.01","input_cache_read":"0.0000000375"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-08-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.4-mini-20260317/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":24.6,"coding_index":56.1,"agentic_index":19.7}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"mistralai/mistral-small-2603","canonical_slug":"mistralai/mistral-small-2603","hugging_face_id":"mistralai/Mistral-Small-4-119B-2603","name":"Mistral: Mistral Small 4","created":1773695685,"description":"Mistral Small 4 is the next major release in the Mistral Small family, unifying the capabilities of several flagship Mistral models into a single system. It combines strong reasoning from...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000006","input_cache_read":"0.000000015"},"top_provider":{"context_length":262144,"max_completion_tokens":209715,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-small-2603/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":11.5,"coding_index":26.6,"agentic_index":1.4}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["high","none"],"default_effort":"high"}},{"id":"mistralai/mistral-small-2603:batch","canonical_slug":"mistralai/mistral-small-2603","hugging_face_id":"mistralai/Mistral-Small-4-119B-2603","name":"Mistral: Mistral Small 4 (batch)","created":1773695685,"description":"Mistral Small 4 is the next major release in the Mistral Small family, unifying the capabilities of several flagship Mistral models into a single system. It combines strong reasoning from...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.000000075","completion":"0.0000003","input_cache_read":"0.0000000075"},"top_provider":{"context_length":262144,"max_completion_tokens":209715,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-small-2603/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":11.5,"coding_index":26.6,"agentic_index":1.4}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["high","none"],"default_effort":"high"}},{"id":"z-ai/glm-5-turbo","canonical_slug":"z-ai/glm-5-turbo-20260315","hugging_face_id":"","name":"Z.ai: GLM 5 Turbo","created":1773583573,"description":"GLM-5 Turbo is a new model from Z.ai designed for fast inference and strong performance in agent-driven environments such as OpenClaw scenarios. It is deeply optimized for real-world agent workflows...","context_length":202752,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000012","completion":"0.000004","input_cache_read":"0.00000024"},"top_provider":{"context_length":202752,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":"2098-12-31","links":{"details":"/api/v1/models/z-ai/glm-5-turbo-20260315/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1278,"win_rate":57,"rank":23},{"arena":"models","category":"asciiart","elo":1171,"win_rate":49.4,"rank":32},{"arena":"models","category":"codecategories","elo":1280,"win_rate":54.8,"rank":25},{"arena":"models","category":"dataviz","elo":1281,"win_rate":55.9,"rank":18},{"arena":"models","category":"gamedev","elo":1275,"win_rate":54.8,"rank":27},{"arena":"models","category":"svg","elo":1232,"win_rate":56,"rank":19},{"arena":"models","category":"uicomponent","elo":1281,"win_rate":56.6,"rank":26},{"arena":"models","category":"website","elo":1278,"win_rate":54,"rank":27}]},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"nvidia/nemotron-3-super-120b-a12b","canonical_slug":"nvidia/nemotron-3-super-120b-a12b-20230311","hugging_face_id":"nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8","name":"NVIDIA: Nemotron 3 Super","created":1773245239,"description":"NVIDIA Nemotron 3 Super is a 120B-parameter open hybrid MoE model, activating just 12B parameters for maximum compute efficiency and accuracy in complex multi-agent applications. Built on a hybrid Mamba-Transformer...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000008","completion":"0.00000045"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nvidia/nemotron-3-super-120b-a12b-20230311/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":13.6,"coding_index":37.7,"agentic_index":4.1}},"reasoning":{"mandatory":false,"default_enabled":true,"supports_max_tokens":true,"supported_efforts":["medium","low"],"default_effort":"medium"}},{"id":"nvidia/nemotron-3-super-120b-a12b:free","canonical_slug":"nvidia/nemotron-3-super-120b-a12b-20230311","hugging_face_id":"nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8","name":"NVIDIA: Nemotron 3 Super (free)","created":1773245239,"description":"NVIDIA Nemotron 3 Super is a 120B-parameter open hybrid MoE model, activating just 12B parameters for maximum compute efficiency and accuracy in complex multi-agent applications. Built on a hybrid Mamba-Transformer...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nvidia/nemotron-3-super-120b-a12b-20230311/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":13.6,"coding_index":37.7,"agentic_index":4.1}},"reasoning":{"mandatory":false,"default_enabled":true,"supports_max_tokens":true,"supported_efforts":["medium","low"],"default_effort":"medium"}},{"id":"bytedance-seed/seed-2.0-lite","canonical_slug":"bytedance-seed/seed-2.0-lite-20260309","hugging_face_id":null,"name":"ByteDance Seed: Seed-2.0-Lite","created":1773157231,"description":"Seed-2.0-Lite is a versatile, cost‑efficient enterprise workhorse that delivers strong multimodal and agent capabilities while offering noticeably lower latency, making it a practical default choice for most production workloads across...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.000002","overrides":[{"min_prompt_tokens":128000,"prompt":"0.0000005","completion":"0.000004"}]},"top_provider":{"context_length":262144,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/bytedance-seed/seed-2.0-lite-20260309/endpoints"},"reasoning":{"mandatory":false,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"qwen/qwen3.5-9b","canonical_slug":"qwen/qwen3.5-9b-20260310","hugging_face_id":"Qwen/Qwen3.5-9B","name":"Qwen: Qwen3.5-9B","created":1773152396,"description":"Qwen3.5-9B is a multimodal foundation model from the Qwen3.5 family, designed to deliver strong reasoning, coding, and visual understanding in an efficient 9B-parameter architecture. It uses a unified vision-language design...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.00000015"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.5-9b-20260310/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":28.7,"agentic_index":null}},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3.5-9b:batch","canonical_slug":"qwen/qwen3.5-9b-20260310","hugging_face_id":"Qwen/Qwen3.5-9B","name":"Qwen: Qwen3.5-9B (batch)","created":1773152396,"description":"Qwen3.5-9B is a multimodal foundation model from the Qwen3.5 family, designed to deliver strong reasoning, coding, and visual understanding in an efficient 9B-parameter architecture. It uses a unified vision-language design...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000017","completion":"0.00000025"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.5-9b-20260310/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":28.7,"agentic_index":null}},"reasoning":{"mandatory":false}},{"id":"openai/gpt-5.4-pro","canonical_slug":"openai/gpt-5.4-pro-20260305","hugging_face_id":"","name":"OpenAI: GPT-5.4 Pro","created":1772734366,"description":"GPT-5.4 Pro is OpenAI's most advanced model, building on GPT-5.4's unified architecture with enhanced reasoning capabilities for complex, high-stakes tasks. It features a 1M+ token context window (922K input, 128K...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00003","completion":"0.00018","web_search":"0.01","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00006","completion":"0.00027"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.4-pro-20260305/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium"],"default_effort":"medium"}},{"id":"openai/gpt-5.4-pro:batch","canonical_slug":"openai/gpt-5.4-pro-20260305","hugging_face_id":"","name":"OpenAI: GPT-5.4 Pro (batch)","created":1772734366,"description":"GPT-5.4 Pro is OpenAI's most advanced model, building on GPT-5.4's unified architecture with enhanced reasoning capabilities for complex, high-stakes tasks. It features a 1M+ token context window (922K input, 128K...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000015","completion":"0.00009","web_search":"0.01","overrides":[{"min_prompt_tokens":272000,"prompt":"0.00003","completion":"0.000135"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.4-pro-20260305/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium"],"default_effort":"medium"}},{"id":"openai/gpt-5.4","canonical_slug":"openai/gpt-5.4-20260305","hugging_face_id":"","name":"OpenAI: GPT-5.4","created":1772734352,"description":"GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window (922K input, 128K output) with support for...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.000015","web_search":"0.01","input_cache_read":"0.00000025","overrides":[{"min_prompt_tokens":272000,"prompt":"0.000005","completion":"0.0000225","input_cache_read":"0.0000005"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.4-20260305/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1130,"win_rate":42.3,"rank":74},{"arena":"models","category":"asciiart","elo":1217,"win_rate":55.4,"rank":22},{"arena":"models","category":"codecategories","elo":1227,"win_rate":52.5,"rank":44},{"arena":"models","category":"dataviz","elo":1251,"win_rate":56.6,"rank":31},{"arena":"models","category":"gamedev","elo":1262,"win_rate":57.6,"rank":30},{"arena":"models","category":"svg","elo":1219,"win_rate":57.8,"rank":22},{"arena":"models","category":"uicomponent","elo":1255,"win_rate":57.4,"rank":33},{"arena":"models","category":"website","elo":1232,"win_rate":52.5,"rank":46},{"arena":"agents","category":"androidnative","elo":1047,"win_rate":47.4,"rank":35},{"arena":"agents","category":"fullstack","elo":1026,"win_rate":40.8,"rank":39},{"arena":"agents","category":"godotgamedev","elo":1135,"win_rate":46.9,"rank":21},{"arena":"agents","category":"mobileapps","elo":1114,"win_rate":46.1,"rank":39},{"arena":"agents","category":"webapps","elo":1072,"win_rate":40.3,"rank":35}],"artificial_analysis":{"intelligence_index":null,"coding_index":71.1,"agentic_index":null}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.4:batch","canonical_slug":"openai/gpt-5.4-20260305","hugging_face_id":"","name":"OpenAI: GPT-5.4 (batch)","created":1772734352,"description":"GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window (922K input, 128K output) with support for...","context_length":1050000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.0000075","web_search":"0.01","input_cache_read":"0.000000125","overrides":[{"min_prompt_tokens":272000,"prompt":"0.0000025","completion":"0.00001125","input_cache_read":"0.00000025"}]},"top_provider":{"context_length":1050000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.4-20260305/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1130,"win_rate":42.3,"rank":74},{"arena":"models","category":"asciiart","elo":1217,"win_rate":55.4,"rank":22},{"arena":"models","category":"codecategories","elo":1227,"win_rate":52.5,"rank":44},{"arena":"models","category":"dataviz","elo":1251,"win_rate":56.6,"rank":31},{"arena":"models","category":"gamedev","elo":1262,"win_rate":57.6,"rank":30},{"arena":"models","category":"svg","elo":1219,"win_rate":57.8,"rank":22},{"arena":"models","category":"uicomponent","elo":1255,"win_rate":57.4,"rank":33},{"arena":"models","category":"website","elo":1232,"win_rate":52.5,"rank":46},{"arena":"agents","category":"androidnative","elo":1047,"win_rate":47.4,"rank":35},{"arena":"agents","category":"fullstack","elo":1026,"win_rate":40.8,"rank":39},{"arena":"agents","category":"godotgamedev","elo":1135,"win_rate":46.9,"rank":21},{"arena":"agents","category":"mobileapps","elo":1114,"win_rate":46.1,"rank":39},{"arena":"agents","category":"webapps","elo":1072,"win_rate":40.3,"rank":35}],"artificial_analysis":{"intelligence_index":null,"coding_index":71.1,"agentic_index":null}},"reasoning":{"mandatory":false,"default_enabled":false,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"inception/mercury-2","canonical_slug":"inception/mercury-2-20260304","hugging_face_id":null,"name":"Inception: Mercury 2","created":1772636275,"description":"Mercury 2 is an extremely fast reasoning LLM, and the first reasoning diffusion LLM (dLLM). Instead of generating tokens sequentially, Mercury 2 produces and refines multiple tokens in parallel, achieving...","context_length":128000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.00000075","input_cache_read":"0.000000025"},"top_provider":{"context_length":128000,"max_completion_tokens":50000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools"],"default_parameters":{"temperature":0.75,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/inception/mercury-2-20260304/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1002,"win_rate":22.3,"rank":104},{"arena":"models","category":"asciiart","elo":1021,"win_rate":26.5,"rank":64},{"arena":"models","category":"codecategories","elo":1026,"win_rate":21.3,"rank":108},{"arena":"models","category":"dataviz","elo":1013,"win_rate":21.8,"rank":104},{"arena":"models","category":"gamedev","elo":1004,"win_rate":19.7,"rank":109},{"arena":"models","category":"svg","elo":1013,"win_rate":24.5,"rank":81},{"arena":"models","category":"uicomponent","elo":997,"win_rate":19.2,"rank":105},{"arena":"models","category":"website","elo":1036,"win_rate":21.6,"rank":112}],"artificial_analysis":{"intelligence_index":11.5,"coding_index":31.1,"agentic_index":4}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","medium","low","none"],"default_effort":"medium"}},{"id":"google/gemini-3.1-flash-lite-preview","canonical_slug":"google/gemini-3.1-flash-lite-preview-20260303","hugging_face_id":"","name":"Google: Gemini 3.1 Flash Lite Preview","created":1772512673,"description":"Gemini 3.1 Flash Lite Preview is Google's high-efficiency model optimized for high-volume use cases. It outperforms Gemini 2.5 Flash Lite on overall quality and approaches Gemini 2.5 Flash performance across...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","video","file","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.0000015","image":"0.00000025","audio":"0.0000005","input_audio_cache":"0.00000005","web_search":"0.014","internal_reasoning":"0.0000015","input_cache_read":"0.000000025","input_cache_write":"0.0000000833333333333333"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.1-flash-lite-preview-20260303/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1076,"win_rate":38.7,"rank":93},{"arena":"models","category":"asciiart","elo":1185,"win_rate":50.7,"rank":27},{"arena":"models","category":"codecategories","elo":1087,"win_rate":36.4,"rank":96},{"arena":"models","category":"dataviz","elo":1061,"win_rate":33.3,"rank":99},{"arena":"models","category":"gamedev","elo":1052,"win_rate":33.7,"rank":101},{"arena":"models","category":"svg","elo":1077,"win_rate":42.5,"rank":66},{"arena":"models","category":"uicomponent","elo":1087,"win_rate":37.7,"rank":91},{"arena":"models","category":"website","elo":1095,"win_rate":36.6,"rank":98}],"artificial_analysis":{"intelligence_index":16,"coding_index":34.7,"agentic_index":3.2}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"minimal"}},{"id":"bytedance-seed/seed-2.0-mini","canonical_slug":"bytedance-seed/seed-2.0-mini-20260224","hugging_face_id":"","name":"ByteDance Seed: Seed-2.0-Mini","created":1772131107,"description":"Seed-2.0-mini targets latency-sensitive, high-concurrency, and cost-sensitive scenarios, emphasizing fast response and flexible inference deployment. It delivers performance comparable to ByteDance-Seed-1.6, supports 256k context, four reasoning effort modes (minimal/low/medium/high), multimodal understanding,...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000004","overrides":[{"min_prompt_tokens":128000,"prompt":"0.0000002","completion":"0.0000008"}]},"top_provider":{"context_length":262144,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/bytedance-seed/seed-2.0-mini-20260224/endpoints"},"reasoning":{"mandatory":false,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"google/gemini-3.1-flash-image-preview","canonical_slug":"google/gemini-3.1-flash-image-preview-20260226","hugging_face_id":"","name":"Google: Nano Banana 2 (Gemini 3.1 Flash Image Preview)","created":1772119558,"description":"Gemini 3.1 Flash Image Preview, a.k.a. \"Nano Banana 2,\" is Google’s latest state of the art image generation and editing model, delivering Pro-level visual quality at Flash speed. It combines...","context_length":65536,"architecture":{"modality":"text+image->text+image","input_modalities":["image","text"],"output_modalities":["image","text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.0000005","completion":"0.000003","image_output":"0.00006","web_search":"0.014"},"top_provider":{"context_length":65536,"max_completion_tokens":58982,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","temperature","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.1-flash-image-preview-20260226/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"graphicdesign","elo":1260,"win_rate":66.3,"rank":4},{"arena":"models","category":"image","elo":1279,"win_rate":65.1,"rank":4},{"arena":"models","category":"logo","elo":1263,"win_rate":62.9,"rank":4}]},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","minimal"],"default_effort":"minimal"}},{"id":"qwen/qwen3.5-35b-a3b","canonical_slug":"qwen/qwen3.5-35b-a3b-20260224","hugging_face_id":"Qwen/Qwen3.5-35B-A3B","name":"Qwen: Qwen3.5-35B-A3B","created":1772053822,"description":"The Qwen3.5 Series 35B-A3B is a native vision-language model designed with a hybrid architecture that integrates linear attention mechanisms and a sparse mixture-of-experts model, achieving higher inference efficiency. Its overall...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.0000003125","completion":"0.00000125","input_cache_read":"0.00000015625"},"top_provider":{"context_length":256000,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":20,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.5-35b-a3b-20260224/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":37,"agentic_index":null}},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3.5-27b","canonical_slug":"qwen/qwen3.5-27b-20260224","hugging_face_id":"Qwen/Qwen3.5-27B","name":"Qwen: Qwen3.5-27B","created":1772053810,"description":"The Qwen3.5 27B native vision-language Dense model incorporates a linear attention mechanism, delivering fast response times while balancing inference speed and performance. Its overall capabilities are comparable to those of...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.000000195","completion":"0.00000156"},"top_provider":{"context_length":262144,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.6,"top_p":0.95,"top_k":20,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.5-27b-20260224/endpoints"},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3.5-122b-a10b","canonical_slug":"qwen/qwen3.5-122b-a10b-20260224","hugging_face_id":"Qwen/Qwen3.5-122B-A10B","name":"Qwen: Qwen3.5-122B-A10B","created":1772053789,"description":"The Qwen3.5 122B-A10B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. In terms of...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000026","completion":"0.00000208"},"top_provider":{"context_length":262144,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.6,"top_p":0.95,"top_k":20,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.5-122b-a10b-20260224/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":16.2,"coding_index":45.7,"agentic_index":9.6}},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3.5-flash-02-23","canonical_slug":"qwen/qwen3.5-flash-20260224","hugging_face_id":null,"name":"Qwen: Qwen3.5-Flash","created":1772053776,"description":"The Qwen3.5 native vision-language Flash models are built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. Compared to the...","context_length":1000000,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.000000065","completion":"0.00000026"},"top_provider":{"context_length":1000000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.5-flash-20260224/endpoints"},"reasoning":{"mandatory":false}},{"id":"google/gemini-3.1-pro-preview-customtools","canonical_slug":"google/gemini-3.1-pro-preview-customtools-20260219","hugging_face_id":null,"name":"Google: Gemini 3.1 Pro Preview Custom Tools","created":1772045923,"description":"Gemini 3.1 Pro Preview Custom Tools is a variant of Gemini 3.1 Pro that improves tool selection behavior by preventing overuse of a general bash tool when more efficient third-party...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","audio","image","video","file"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000012","image":"0.000002","audio":"0.000002","input_audio_cache":"0.0000002","web_search":"0.014","internal_reasoning":"0.000012","input_cache_read":"0.0000002","input_cache_write":"0.000000375","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000004","completion":"0.000018","audio":"0.000004","input_audio_cache":"0.0000004","input_cache_read":"0.0000004"}]},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.1-pro-preview-customtools-20260219/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"openai/gpt-5.3-codex","canonical_slug":"openai/gpt-5.3-codex-20260224","hugging_face_id":"","name":"OpenAI: GPT-5.3-Codex","created":1771959164,"description":"GPT-5.3-Codex is OpenAI’s most advanced agentic coding model, combining the frontier software engineering performance of GPT-5.2-Codex with the broader reasoning and professional knowledge capabilities of GPT-5.2. It achieves state-of-the-art results...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000175","completion":"0.000014","web_search":"0.01","input_cache_read":"0.000000175"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.3-codex-20260224/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"androidnative","elo":1084,"win_rate":35.2,"rank":30},{"arena":"agents","category":"fullstack","elo":992,"win_rate":36.5,"rank":42},{"arena":"agents","category":"godotgamedev","elo":1124,"win_rate":45.1,"rank":23},{"arena":"agents","category":"mobileapps","elo":1081,"win_rate":41.4,"rank":41},{"arena":"agents","category":"webapps","elo":1046,"win_rate":36.9,"rank":37},{"arena":"models","category":"3d","elo":1036,"win_rate":35.3,"rank":98},{"arena":"models","category":"asciiart","elo":1169,"win_rate":51.2,"rank":34},{"arena":"models","category":"codecategories","elo":1159,"win_rate":47.2,"rank":72},{"arena":"models","category":"dataviz","elo":1180,"win_rate":50.6,"rank":64},{"arena":"models","category":"gamedev","elo":1184,"win_rate":51.3,"rank":58},{"arena":"models","category":"svg","elo":1154,"win_rate":54,"rank":48},{"arena":"models","category":"uicomponent","elo":1153,"win_rate":47.3,"rank":71},{"arena":"models","category":"website","elo":1175,"win_rate":48.6,"rank":71}]},"reasoning":{"mandatory":false,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"aion-labs/aion-2.0","canonical_slug":"aion-labs/aion-2.0-20260223","hugging_face_id":null,"name":"AionLabs: Aion-2.0","created":1771881306,"description":"Aion-2.0 is a variant of DeepSeek V3.2 optimized for immersive roleplaying and storytelling. It is particularly strong at introducing tension, crises, and conflict into stories, making narratives feel more engaging....","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000008","completion":"0.0000016","input_cache_read":"0.0000002"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/aion-labs/aion-2.0-20260223/endpoints"},"reasoning":{"mandatory":true}},{"id":"google/gemini-3.1-pro-preview","canonical_slug":"google/gemini-3.1-pro-preview-20260219","hugging_face_id":"","name":"Google: Gemini 3.1 Pro Preview","created":1771509627,"description":"Gemini 3.1 Pro Preview is Google’s frontier reasoning model, delivering enhanced software engineering performance, improved agentic reliability, and more efficient token usage across complex workflows. Building on the multimodal foundation...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["audio","file","image","text","video"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000012","image":"0.000002","audio":"0.000002","input_audio_cache":"0.0000002","web_search":"0.014","internal_reasoning":"0.000012","input_cache_read":"0.0000002","input_cache_write":"0.000000375","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000004","completion":"0.000018","audio":"0.000004","input_audio_cache":"0.0000004","input_cache_read":"0.0000004"}]},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.1-pro-preview-20260219/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1114,"win_rate":44.2,"rank":22},{"arena":"agents","category":"agentichtmlslides","elo":1226,"win_rate":55.8,"rank":4},{"arena":"agents","category":"agenticslides","elo":1112,"win_rate":33.8,"rank":7},{"arena":"agents","category":"agenticslides(html)","elo":1219,"win_rate":54.4,"rank":4},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1107,"win_rate":33.9,"rank":7},{"arena":"agents","category":"androidnative","elo":1065,"win_rate":41.3,"rank":33},{"arena":"agents","category":"fullstack","elo":1078,"win_rate":41.5,"rank":31},{"arena":"agents","category":"godotgamedev","elo":1236,"win_rate":60,"rank":5},{"arena":"agents","category":"htmlslides","elo":1155,"win_rate":48.6,"rank":16},{"arena":"agents","category":"mobileapps","elo":1119,"win_rate":43.6,"rank":35},{"arena":"agents","category":"pptxslides","elo":1110,"win_rate":34.1,"rank":7},{"arena":"agents","category":"python-pptxslides","elo":1109,"win_rate":31.9,"rank":25},{"arena":"agents","category":"webapps","elo":1139,"win_rate":44.9,"rank":28},{"arena":"models","category":"3d","elo":1263,"win_rate":58.5,"rank":30},{"arena":"models","category":"asciiart","elo":1288,"win_rate":63.5,"rank":10},{"arena":"models","category":"codecategories","elo":1259,"win_rate":64.4,"rank":33},{"arena":"models","category":"dataviz","elo":1255,"win_rate":61.8,"rank":29},{"arena":"models","category":"gamedev","elo":1226,"win_rate":52.3,"rank":42},{"arena":"models","category":"svg","elo":1311,"win_rate":67.8,"rank":8},{"arena":"models","category":"uicomponent","elo":1295,"win_rate":69.4,"rank":18},{"arena":"models","category":"website","elo":1265,"win_rate":64.4,"rank":32}],"artificial_analysis":{"intelligence_index":30.4,"coding_index":68.8,"agentic_index":10.3}},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"google/gemini-3.1-pro-preview:batch","canonical_slug":"google/gemini-3.1-pro-preview-20260219","hugging_face_id":"","name":"Google: Gemini 3.1 Pro Preview (batch)","created":1771509627,"description":"Gemini 3.1 Pro Preview is Google’s frontier reasoning model, delivering enhanced software engineering performance, improved agentic reliability, and more efficient token usage across complex workflows. Building on the multimodal foundation...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["audio","file","image","text","video"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000006","image":"0.000001","audio":"0.000001","web_search":"0.014","internal_reasoning":"0.000006","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000002","completion":"0.000009","audio":"0.000002"}]},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3.1-pro-preview-20260219/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1114,"win_rate":44.2,"rank":22},{"arena":"agents","category":"agentichtmlslides","elo":1226,"win_rate":55.8,"rank":4},{"arena":"agents","category":"agenticslides","elo":1112,"win_rate":33.8,"rank":7},{"arena":"agents","category":"agenticslides(html)","elo":1219,"win_rate":54.4,"rank":4},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1107,"win_rate":33.9,"rank":7},{"arena":"agents","category":"androidnative","elo":1065,"win_rate":41.3,"rank":33},{"arena":"agents","category":"fullstack","elo":1078,"win_rate":41.5,"rank":31},{"arena":"agents","category":"godotgamedev","elo":1236,"win_rate":60,"rank":5},{"arena":"agents","category":"htmlslides","elo":1155,"win_rate":48.6,"rank":16},{"arena":"agents","category":"mobileapps","elo":1119,"win_rate":43.6,"rank":35},{"arena":"agents","category":"pptxslides","elo":1110,"win_rate":34.1,"rank":7},{"arena":"agents","category":"python-pptxslides","elo":1109,"win_rate":31.9,"rank":25},{"arena":"agents","category":"webapps","elo":1139,"win_rate":44.9,"rank":28},{"arena":"models","category":"3d","elo":1263,"win_rate":58.5,"rank":30},{"arena":"models","category":"asciiart","elo":1288,"win_rate":63.5,"rank":10},{"arena":"models","category":"codecategories","elo":1259,"win_rate":64.4,"rank":33},{"arena":"models","category":"dataviz","elo":1255,"win_rate":61.8,"rank":29},{"arena":"models","category":"gamedev","elo":1226,"win_rate":52.3,"rank":42},{"arena":"models","category":"svg","elo":1311,"win_rate":67.8,"rank":8},{"arena":"models","category":"uicomponent","elo":1295,"win_rate":69.4,"rank":18},{"arena":"models","category":"website","elo":1265,"win_rate":64.4,"rank":32}],"artificial_analysis":{"intelligence_index":30.4,"coding_index":68.8,"agentic_index":10.3}},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"anthropic/claude-sonnet-4.6","canonical_slug":"anthropic/claude-4.6-sonnet-20260217","hugging_face_id":"","name":"Anthropic: Claude Sonnet 4.6","created":1771342990,"description":"Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It excels at iterative development, complex codebase navigation, end-to-end project management with...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000003","completion":"0.000015","web_search":"0.01","input_cache_read":"0.0000003","input_cache_write":"0.00000375","input_cache_write_1h":"0.000006"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.6-sonnet-20260217/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1177,"win_rate":52.2,"rank":16},{"arena":"agents","category":"androidnative","elo":1212,"win_rate":62,"rank":13},{"arena":"agents","category":"fullstack","elo":1222,"win_rate":64.3,"rank":15},{"arena":"agents","category":"godotgamedev","elo":1226,"win_rate":60.6,"rank":6},{"arena":"agents","category":"mobileapps","elo":1239,"win_rate":63.6,"rank":11},{"arena":"agents","category":"webapps","elo":1203,"win_rate":56.9,"rank":21},{"arena":"models","category":"3d","elo":1263,"win_rate":57.6,"rank":29},{"arena":"models","category":"asciiart","elo":1252,"win_rate":60.1,"rank":16},{"arena":"models","category":"codecategories","elo":1291,"win_rate":59.4,"rank":18},{"arena":"models","category":"dataviz","elo":1295,"win_rate":58.2,"rank":13},{"arena":"models","category":"gamedev","elo":1282,"win_rate":58.8,"rank":25},{"arena":"models","category":"svg","elo":1222,"win_rate":58.7,"rank":20},{"arena":"models","category":"uicomponent","elo":1287,"win_rate":58.3,"rank":21},{"arena":"models","category":"website","elo":1297,"win_rate":60,"rank":14}],"artificial_analysis":{"intelligence_index":30.5,"coding_index":63,"agentic_index":33.1}},"reasoning":{"mandatory":false,"supported_efforts":["max","high","medium","low"],"default_effort":"medium"}},{"id":"anthropic/claude-sonnet-4.6:batch","canonical_slug":"anthropic/claude-4.6-sonnet-20260217","hugging_face_id":"","name":"Anthropic: Claude Sonnet 4.6 (batch)","created":1771342990,"description":"Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It excels at iterative development, complex codebase navigation, end-to-end project management with...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.0000015","completion":"0.0000075","web_search":"0.01","input_cache_read":"0.00000015","input_cache_write":"0.000001875","input_cache_write_1h":"0.000003"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_p","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.6-sonnet-20260217/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1177,"win_rate":52.2,"rank":16},{"arena":"agents","category":"androidnative","elo":1212,"win_rate":62,"rank":13},{"arena":"agents","category":"fullstack","elo":1222,"win_rate":64.3,"rank":15},{"arena":"agents","category":"godotgamedev","elo":1226,"win_rate":60.6,"rank":6},{"arena":"agents","category":"mobileapps","elo":1239,"win_rate":63.6,"rank":11},{"arena":"agents","category":"webapps","elo":1203,"win_rate":56.9,"rank":21},{"arena":"models","category":"3d","elo":1263,"win_rate":57.6,"rank":29},{"arena":"models","category":"asciiart","elo":1252,"win_rate":60.1,"rank":16},{"arena":"models","category":"codecategories","elo":1291,"win_rate":59.4,"rank":18},{"arena":"models","category":"dataviz","elo":1295,"win_rate":58.2,"rank":13},{"arena":"models","category":"gamedev","elo":1282,"win_rate":58.8,"rank":25},{"arena":"models","category":"svg","elo":1222,"win_rate":58.7,"rank":20},{"arena":"models","category":"uicomponent","elo":1287,"win_rate":58.3,"rank":21},{"arena":"models","category":"website","elo":1297,"win_rate":60,"rank":14}],"artificial_analysis":{"intelligence_index":30.5,"coding_index":63,"agentic_index":33.1}},"reasoning":{"mandatory":false,"supported_efforts":["max","high","medium","low"],"default_effort":"medium"}},{"id":"qwen/qwen3.5-plus-02-15","canonical_slug":"qwen/qwen3.5-plus-20260216","hugging_face_id":"","name":"Qwen: Qwen3.5 Plus 2026-02-15","created":1771229416,"description":"The Qwen3.5 native vision-language series Plus models are built on a hybrid architecture that integrates linear attention mechanisms with sparse mixture-of-experts models, achieving higher inference efficiency. In a variety of...","context_length":1000000,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000026","completion":"0.00000156","overrides":[{"min_prompt_tokens":256000,"prompt":"0.000000325","completion":"0.00000195"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.5-plus-20260216/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1147,"win_rate":47.7,"rank":69},{"arena":"models","category":"asciiart","elo":1112,"win_rate":43.2,"rank":57},{"arena":"models","category":"codecategories","elo":1182,"win_rate":48.5,"rank":65},{"arena":"models","category":"dataviz","elo":1152,"win_rate":44.9,"rank":76},{"arena":"models","category":"gamedev","elo":1128,"win_rate":42.6,"rank":79},{"arena":"models","category":"svg","elo":1132,"win_rate":48.9,"rank":54},{"arena":"models","category":"uicomponent","elo":1193,"win_rate":52.2,"rank":56},{"arena":"models","category":"website","elo":1200,"win_rate":50,"rank":58}]},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3.5-397b-a17b","canonical_slug":"qwen/qwen3.5-397b-a17b-20260216","hugging_face_id":"Qwen/Qwen3.5-397B-A17B","name":"Qwen: Qwen3.5 397B A17B","created":1771223018,"description":"The Qwen3.5 series 397B-A17B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. It delivers...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["text","image","video"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000055","completion":"0.0000035","input_cache_read":"0.000000225"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.6,"top_p":0.95,"top_k":20,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3.5-397b-a17b-20260216/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1189,"win_rate":56.7,"rank":52},{"arena":"models","category":"codecategories","elo":1195,"win_rate":52.6,"rank":55},{"arena":"models","category":"dataviz","elo":1191,"win_rate":53.2,"rank":58},{"arena":"models","category":"gamedev","elo":1163,"win_rate":50.1,"rank":64},{"arena":"models","category":"svg","elo":1164,"win_rate":56.1,"rank":45},{"arena":"models","category":"uicomponent","elo":1179,"win_rate":51.4,"rank":64},{"arena":"models","category":"website","elo":1203,"win_rate":52.5,"rank":56}],"artificial_analysis":{"intelligence_index":19.1,"coding_index":48.2,"agentic_index":10.6}},"reasoning":{"mandatory":false}},{"id":"minimax/minimax-m2.5","canonical_slug":"minimax/minimax-m2.5-20260211","hugging_face_id":"MiniMaxAI/MiniMax-M2.5","name":"MiniMax: MiniMax M2.5","created":1770908502,"description":"MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex real-world digital working environments, M2.5 builds upon the coding expertise of M2.1...","context_length":204800,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000027","completion":"0.00000108","input_cache_read":"0.000000027"},"top_provider":{"context_length":200000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/minimax/minimax-m2.5-20260211/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1198,"win_rate":57.6,"rank":49},{"arena":"models","category":"codecategories","elo":1222,"win_rate":56.8,"rank":46},{"arena":"models","category":"dataviz","elo":1188,"win_rate":51.2,"rank":59},{"arena":"models","category":"gamedev","elo":1199,"win_rate":55.5,"rank":50},{"arena":"models","category":"svg","elo":1174,"win_rate":54.5,"rank":40},{"arena":"models","category":"uicomponent","elo":1188,"win_rate":53.4,"rank":58},{"arena":"models","category":"website","elo":1234,"win_rate":57.5,"rank":44}]},"reasoning":{"mandatory":true}},{"id":"z-ai/glm-5","canonical_slug":"z-ai/glm-5-20260211","hugging_face_id":"zai-org/GLM-5","name":"Z.ai: GLM 5","created":1770829182,"description":"GLM-5 is Z.ai’s flagship open-source foundation model engineered for complex systems design and long-horizon agent workflows. Built for expert developers, it delivers production-grade performance on large-scale programming tasks, rivaling leading...","context_length":204800,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000006","completion":"0.00000192","input_cache_read":"0.00000012"},"top_provider":{"context_length":198000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-5-20260211/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"androidnative","elo":1161,"win_rate":55.2,"rank":24},{"arena":"agents","category":"fullstack","elo":1131,"win_rate":51.7,"rank":27},{"arena":"agents","category":"godotgamedev","elo":1144,"win_rate":46.7,"rank":15},{"arena":"agents","category":"htmlslides","elo":1150,"win_rate":45.1,"rank":18},{"arena":"agents","category":"mobileapps","elo":1163,"win_rate":51.6,"rank":29},{"arena":"models","category":"3d","elo":1257,"win_rate":56.3,"rank":31},{"arena":"models","category":"asciiart","elo":1165,"win_rate":47.9,"rank":37},{"arena":"models","category":"codecategories","elo":1261,"win_rate":55.5,"rank":32},{"arena":"models","category":"dataviz","elo":1244,"win_rate":53,"rank":36},{"arena":"models","category":"gamedev","elo":1257,"win_rate":57.4,"rank":31},{"arena":"models","category":"svg","elo":1193,"win_rate":54.4,"rank":33},{"arena":"models","category":"uicomponent","elo":1246,"win_rate":53.6,"rank":39},{"arena":"models","category":"website","elo":1260,"win_rate":55,"rank":34}]},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"qwen/qwen3-max-thinking","canonical_slug":"qwen/qwen3-max-thinking-20260123","hugging_face_id":null,"name":"Qwen: Qwen3 Max Thinking","created":1770671901,"description":"Qwen3-Max-Thinking is the flagship reasoning model in the Qwen3 series, designed for high-stakes cognitive tasks that require deep, multi-step reasoning. By significantly scaling model capacity and reinforcement learning compute, it...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.00000078","completion":"0.0000039","overrides":[{"min_prompt_tokens":32000,"prompt":"0.00000156","completion":"0.0000078"},{"min_prompt_tokens":128000,"prompt":"0.00000195","completion":"0.00000975"}]},"top_provider":{"context_length":262144,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-max-thinking-20260123/endpoints"},"reasoning":{"mandatory":false}},{"id":"anthropic/claude-opus-4.6","canonical_slug":"anthropic/claude-4.6-opus-20260205","hugging_face_id":"","name":"Anthropic: Claude Opus 4.6","created":1770219050,"description":"Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire workflows rather than single prompts, making it especially effective...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000025","web_search":"0.01","input_cache_read":"0.0000005","input_cache_write":"0.00000625","input_cache_write_1h":"0.00001"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.6-opus-20260205/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1215,"win_rate":50.3,"rank":8},{"arena":"agents","category":"androidnative","elo":1219,"win_rate":64.7,"rank":11},{"arena":"agents","category":"fullstack","elo":1228,"win_rate":61.1,"rank":13},{"arena":"agents","category":"mobileapps","elo":1236,"win_rate":60.3,"rank":12},{"arena":"agents","category":"webapps","elo":1217,"win_rate":57.3,"rank":16},{"arena":"models","category":"3d","elo":1303,"win_rate":62.5,"rank":15},{"arena":"models","category":"asciiart","elo":1272,"win_rate":61.9,"rank":13},{"arena":"models","category":"codecategories","elo":1304,"win_rate":61.2,"rank":14},{"arena":"models","category":"dataviz","elo":1297,"win_rate":58.7,"rank":12},{"arena":"models","category":"gamedev","elo":1301,"win_rate":61,"rank":15},{"arena":"models","category":"svg","elo":1253,"win_rate":60.9,"rank":13},{"arena":"models","category":"uicomponent","elo":1300,"win_rate":59.8,"rank":16},{"arena":"models","category":"website","elo":1303,"win_rate":61.2,"rank":13}]},"reasoning":{"mandatory":false,"default_enabled":false,"supports_max_tokens":true,"supported_efforts":["max","high","medium","low"],"default_effort":"high"}},{"id":"anthropic/claude-opus-4.6:batch","canonical_slug":"anthropic/claude-4.6-opus-20260205","hugging_face_id":"","name":"Anthropic: Claude Opus 4.6 (batch)","created":1770219050,"description":"Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire workflows rather than single prompts, making it especially effective...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.0000125","web_search":"0.01","input_cache_read":"0.00000025","input_cache_write":"0.000003125","input_cache_write_1h":"0.000005"},"top_provider":{"context_length":1000000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_p","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.6-opus-20260205/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticgamedev","elo":1215,"win_rate":50.3,"rank":8},{"arena":"agents","category":"androidnative","elo":1219,"win_rate":64.7,"rank":11},{"arena":"agents","category":"fullstack","elo":1228,"win_rate":61.1,"rank":13},{"arena":"agents","category":"mobileapps","elo":1236,"win_rate":60.3,"rank":12},{"arena":"agents","category":"webapps","elo":1217,"win_rate":57.3,"rank":16},{"arena":"models","category":"3d","elo":1303,"win_rate":62.5,"rank":15},{"arena":"models","category":"asciiart","elo":1272,"win_rate":61.9,"rank":13},{"arena":"models","category":"codecategories","elo":1304,"win_rate":61.2,"rank":14},{"arena":"models","category":"dataviz","elo":1297,"win_rate":58.7,"rank":12},{"arena":"models","category":"gamedev","elo":1301,"win_rate":61,"rank":15},{"arena":"models","category":"svg","elo":1253,"win_rate":60.9,"rank":13},{"arena":"models","category":"uicomponent","elo":1300,"win_rate":59.8,"rank":16},{"arena":"models","category":"website","elo":1303,"win_rate":61.2,"rank":13}]},"reasoning":{"mandatory":false,"default_enabled":false,"supports_max_tokens":true,"supported_efforts":["max","high","medium","low"],"default_effort":"high"}},{"id":"qwen/qwen3-coder-next","canonical_slug":"qwen/qwen3-coder-next-2025-02-03","hugging_face_id":"Qwen/Qwen3-Coder-Next","name":"Qwen: Qwen3 Coder Next","created":1770164101,"description":"Qwen3-Coder-Next is an open-weight causal language model optimized for coding agents and local development workflows. It uses a sparse MoE design with 80B total parameters and only 3B activated per...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.00000012","completion":"0.0000008","input_cache_read":"0.00000007"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-coder-next-2025-02-03/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":10.1,"coding_index":36.2,"agentic_index":3.6}}},{"id":"openrouter/free","canonical_slug":"openrouter/free","hugging_face_id":"","name":"Free Models Router","created":1769917427,"description":"The simplest way to get free inference. openrouter/free is a router that selects free models at random from the models available on OpenRouter. The router smartly filters for models that...","context_length":200000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"0","completion":"0"},"top_provider":{"context_length":null,"max_completion_tokens":null,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_completion_tokens","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openrouter/free/endpoints"}},{"id":"stepfun/step-3.5-flash","canonical_slug":"stepfun/step-3.5-flash","hugging_face_id":"stepfun-ai/Step-3.5-Flash","name":"StepFun: Step 3.5 Flash","created":1769728337,"description":"Step 3.5 Flash is StepFun's most capable open-source foundation model. Built on a sparse Mixture of Experts (MoE) architecture, it selectively activates only 11B of its 196B parameters per token....","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000003"},"top_provider":{"context_length":262144,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","reasoning","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/stepfun/step-3.5-flash/endpoints"},"reasoning":{"mandatory":true}},{"id":"moonshotai/kimi-k2.5","canonical_slug":"moonshotai/kimi-k2.5-0127","hugging_face_id":"moonshotai/Kimi-K2.5","name":"MoonshotAI: Kimi K2.5","created":1769487076,"description":"Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-directed agent swarm paradigm. Built on Kimi K2 with continued pretraining over approximately 15T mixed...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000045","completion":"0.00000225","input_cache_read":"0.00000007"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/moonshotai/kimi-k2.5-0127/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"androidnative","elo":1134,"win_rate":57.8,"rank":26},{"arena":"agents","category":"fullstack","elo":1122,"win_rate":54.5,"rank":28},{"arena":"agents","category":"godotgamedev","elo":1219,"win_rate":59.8,"rank":8},{"arena":"agents","category":"mobileapps","elo":1174,"win_rate":54.6,"rank":24},{"arena":"agents","category":"webapps","elo":1150,"win_rate":51.6,"rank":26},{"arena":"models","category":"3d","elo":1233,"win_rate":53.1,"rank":38},{"arena":"models","category":"asciiart","elo":1185,"win_rate":46.4,"rank":28},{"arena":"models","category":"codecategories","elo":1251,"win_rate":54.1,"rank":36},{"arena":"models","category":"dataviz","elo":1235,"win_rate":51.3,"rank":41},{"arena":"models","category":"gamedev","elo":1229,"win_rate":53.4,"rank":41},{"arena":"models","category":"svg","elo":1175,"win_rate":48.4,"rank":39},{"arena":"models","category":"uicomponent","elo":1252,"win_rate":53.6,"rank":37},{"arena":"models","category":"website","elo":1261,"win_rate":55.2,"rank":33}],"artificial_analysis":{"intelligence_index":null,"coding_index":46.8,"agentic_index":null}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"upstage/solar-pro-3","canonical_slug":"upstage/solar-pro-3","hugging_face_id":"","name":"Upstage: Solar Pro 3","created":1769481200,"description":"Solar Pro 3 is Upstage's powerful Mixture-of-Experts (MoE) language model. With 102B total parameters and 12B active parameters per forward pass, it delivers exceptional performance while maintaining computational efficiency. Optimized...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000006","input_cache_read":"0.000000015"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","parallel_tool_calls","presence_penalty","reasoning","response_format","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/upstage/solar-pro-3/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":7.8,"coding_index":16.2,"agentic_index":1.4}},"reasoning":{"mandatory":false}},{"id":"minimax/minimax-m2-her","canonical_slug":"minimax/minimax-m2-her-20260123","hugging_face_id":"","name":"MiniMax: MiniMax M2-her","created":1769177239,"description":"MiniMax M2-her is a dialogue-first large language model built for immersive roleplay, character-driven chat, and expressive multi-turn conversations. Designed to stay consistent in tone and personality, it supports rich message...","context_length":65536,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000012","input_cache_read":"0.00000003"},"top_provider":{"context_length":65536,"max_completion_tokens":2048,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_tokens","temperature","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/minimax/minimax-m2-her-20260123/endpoints"}},{"id":"writer/palmyra-x5","canonical_slug":"writer/palmyra-x5-20250428","hugging_face_id":"","name":"Writer: Palmyra X5","created":1769003823,"description":"Palmyra X5 is Writer's most advanced model, purpose-built for building and scaling AI agents across the enterprise. It delivers industry-leading speed and efficiency on context windows up to 1 million...","context_length":1040000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000006","completion":"0.000006"},"top_provider":{"context_length":1040000,"max_completion_tokens":8192,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_tokens","stop","temperature","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/writer/palmyra-x5-20250428/endpoints"}},{"id":"openai/gpt-audio","canonical_slug":"openai/gpt-audio","hugging_face_id":"","name":"OpenAI: GPT Audio","created":1768862569,"description":"The gpt-audio model is OpenAI's first generally available audio model. The new snapshot features an upgraded decoder for more natural sounding voices and maintains better voice consistency. Audio is priced...","context_length":128000,"architecture":{"modality":"text+audio->text+audio","input_modalities":["text","audio"],"output_modalities":["text","audio"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.00001","audio":"0.000032","audio_output":"0.000064"},"top_provider":{"context_length":128000,"max_completion_tokens":16384,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-audio/endpoints"}},{"id":"openai/gpt-audio-mini","canonical_slug":"openai/gpt-audio-mini","hugging_face_id":"","name":"OpenAI: GPT Audio Mini","created":1768859419,"description":"A cost-efficient version of GPT Audio. The new snapshot features an upgraded decoder for more natural sounding voices and maintains better voice consistency. Input is priced at $0.60 per million...","context_length":128000,"architecture":{"modality":"text+audio->text+audio","input_modalities":["text","audio"],"output_modalities":["text","audio"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000006","completion":"0.0000024","audio":"0.0000006","audio_output":"0.0000024"},"top_provider":{"context_length":128000,"max_completion_tokens":16384,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-audio-mini/endpoints"}},{"id":"z-ai/glm-4.7-flash","canonical_slug":"z-ai/glm-4.7-flash-20260119","hugging_face_id":"zai-org/GLM-4.7-Flash","name":"Z.ai: GLM 4.7 Flash","created":1768833913,"description":"As a 30B-class SOTA model, GLM-4.7-Flash offers a new option that balances performance and efficiency. It is further optimized for agentic coding use cases, strengthening coding capabilities, long-horizon task planning,...","context_length":200000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000000605","completion":"0.0000004"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-4.7-flash-20260119/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1149,"win_rate":51.2,"rank":68},{"arena":"models","category":"codecategories","elo":1192,"win_rate":53.1,"rank":57},{"arena":"models","category":"dataviz","elo":1139,"win_rate":45.3,"rank":83},{"arena":"models","category":"gamedev","elo":1158,"win_rate":49.7,"rank":68},{"arena":"models","category":"svg","elo":1062,"win_rate":44.2,"rank":69},{"arena":"models","category":"uicomponent","elo":1223,"win_rate":57.6,"rank":45},{"arena":"models","category":"website","elo":1206,"win_rate":54,"rank":53}]},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"openai/gpt-5.2-codex","canonical_slug":"openai/gpt-5.2-codex-20260114","hugging_face_id":"","name":"OpenAI: GPT-5.2-Codex","created":1768409315,"description":"GPT-5.2-Codex is an upgraded version of GPT-5.1-Codex optimized for software engineering and coding workflows. It is designed for both interactive development sessions and long, independent execution of complex engineering tasks....","context_length":400000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000175","completion":"0.000014","web_search":"0.01","input_cache_read":"0.000000175"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.2-codex-20260114/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"androidnative","elo":1176,"win_rate":47.5,"rank":20},{"arena":"agents","category":"fullstack","elo":1001,"win_rate":37.2,"rank":41},{"arena":"agents","category":"godotgamedev","elo":1142,"win_rate":47.8,"rank":18},{"arena":"agents","category":"mobileapps","elo":1117,"win_rate":46.6,"rank":37},{"arena":"agents","category":"webapps","elo":1064,"win_rate":39.7,"rank":36}]},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium","low"],"default_effort":"medium"}},{"id":"bytedance-seed/seed-1.6-flash","canonical_slug":"bytedance-seed/seed-1.6-flash-20250625","hugging_face_id":"","name":"ByteDance Seed: Seed 1.6 Flash","created":1766505011,"description":"Seed 1.6 Flash is an ultra-fast multimodal deep thinking model by ByteDance Seed, supporting both text and visual understanding. It features a 256k context window and can generate outputs of...","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["image","text","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000075","completion":"0.0000003","overrides":[{"min_prompt_tokens":128000,"prompt":"0.0000001","completion":"0.0000008"}]},"top_provider":{"context_length":262144,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","reasoning","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/bytedance-seed/seed-1.6-flash-20250625/endpoints"},"reasoning":{"mandatory":false}},{"id":"bytedance-seed/seed-1.6","canonical_slug":"bytedance-seed/seed-1.6-20250625","hugging_face_id":"","name":"ByteDance Seed: Seed 1.6","created":1766504997,"description":"Seed 1.6 is a general-purpose model released by the ByteDance Seed team. It incorporates multimodal capabilities and adaptive deep thinking with a 256K context window.","context_length":262144,"architecture":{"modality":"text+image+video->text","input_modalities":["image","text","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.000002","overrides":[{"min_prompt_tokens":128000,"prompt":"0.0000005","completion":"0.000004"}]},"top_provider":{"context_length":262144,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","reasoning","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/bytedance-seed/seed-1.6-20250625/endpoints"},"reasoning":{"mandatory":false}},{"id":"minimax/minimax-m2.1","canonical_slug":"minimax/minimax-m2.1","hugging_face_id":"MiniMaxAI/MiniMax-M2.1","name":"MiniMax: MiniMax M2.1","created":1766454997,"description":"MiniMax-M2.1 is a lightweight, state-of-the-art large language model optimized for coding, agentic workflows, and modern application development. With only 10 billion activated parameters, it delivers a major jump in real-world...","context_length":204800,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000012","input_cache_read":"0.00000003"},"top_provider":{"context_length":204800,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":1,"top_p":0.9,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/minimax/minimax-m2.1/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1193,"win_rate":57.5,"rank":51},{"arena":"models","category":"codecategories","elo":1205,"win_rate":55.3,"rank":50},{"arena":"models","category":"dataviz","elo":1222,"win_rate":57,"rank":45},{"arena":"models","category":"gamedev","elo":1157,"win_rate":50.4,"rank":69},{"arena":"models","category":"svg","elo":1156,"win_rate":55.4,"rank":47},{"arena":"models","category":"uicomponent","elo":1238,"win_rate":60.9,"rank":41},{"arena":"models","category":"website","elo":1213,"win_rate":55.4,"rank":49}]},"reasoning":{"mandatory":true}},{"id":"z-ai/glm-4.7","canonical_slug":"z-ai/glm-4.7-20251222","hugging_face_id":"zai-org/GLM-4.7","name":"Z.ai: GLM 4.7","created":1766378014,"description":"GLM-4.7 is Z.ai’s latest flagship model, featuring upgrades in two key areas: enhanced programming capabilities and more stable multi-step reasoning/execution. It demonstrates significant improvements in executing complex agent tasks while...","context_length":204800,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000004","completion":"0.00000175","input_cache_read":"0.00000008"},"top_provider":{"context_length":202752,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_a","top_k","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-4.7-20251222/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"androidnative","elo":1129,"win_rate":56.2,"rank":27},{"arena":"agents","category":"fullstack","elo":1058,"win_rate":45.1,"rank":36},{"arena":"agents","category":"godotgamedev","elo":1058,"win_rate":35.8,"rank":27},{"arena":"agents","category":"mobileapps","elo":1132,"win_rate":48.9,"rank":31},{"arena":"models","category":"3d","elo":1219,"win_rate":54.3,"rank":43},{"arena":"models","category":"asciiart","elo":1183,"win_rate":48.1,"rank":30},{"arena":"models","category":"codecategories","elo":1232,"win_rate":54.8,"rank":43},{"arena":"models","category":"dataviz","elo":1212,"win_rate":51.2,"rank":50},{"arena":"models","category":"gamedev","elo":1211,"win_rate":55.2,"rank":46},{"arena":"models","category":"svg","elo":1168,"win_rate":54.3,"rank":42},{"arena":"models","category":"uicomponent","elo":1215,"win_rate":51.1,"rank":46},{"arena":"models","category":"website","elo":1238,"win_rate":55.3,"rank":42}],"artificial_analysis":{"intelligence_index":null,"coding_index":45.3,"agentic_index":null}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"google/gemini-3-flash-preview","canonical_slug":"google/gemini-3-flash-preview-20251217","hugging_face_id":"","name":"Google: Gemini 3 Flash Preview","created":1765987078,"description":"Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance. It delivers near Pro level reasoning and tool...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","file","audio","video"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.0000005","completion":"0.000003","image":"0.0000005","audio":"0.000001","input_audio_cache":"0.0000001","web_search":"0.014","internal_reasoning":"0.000003","input_cache_read":"0.00000005","input_cache_write":"0.0000000833333333333333"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3-flash-preview-20251217/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticslides","elo":1073,"win_rate":39.3,"rank":8},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1075,"win_rate":39.3,"rank":8},{"arena":"agents","category":"androidnative","elo":1066,"win_rate":48.1,"rank":32},{"arena":"agents","category":"fullstack","elo":1072,"win_rate":47.4,"rank":33},{"arena":"agents","category":"godotgamedev","elo":1161,"win_rate":50.6,"rank":13},{"arena":"agents","category":"mobileapps","elo":1117,"win_rate":46.6,"rank":36},{"arena":"agents","category":"python-pptxslides","elo":1036,"win_rate":38.3,"rank":27},{"arena":"agents","category":"webapps","elo":1124,"win_rate":48.5,"rank":31},{"arena":"models","category":"3d","elo":1212,"win_rate":62.7,"rank":45},{"arena":"models","category":"codecategories","elo":1204,"win_rate":57.6,"rank":51},{"arena":"models","category":"gamedev","elo":1189,"win_rate":58.3,"rank":57},{"arena":"models","category":"website","elo":1208,"win_rate":57,"rank":51}]},"reasoning":{"mandatory":false,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"google/gemini-3-flash-preview:batch","canonical_slug":"google/gemini-3-flash-preview-20251217","hugging_face_id":"","name":"Google: Gemini 3 Flash Preview (batch)","created":1765987078,"description":"Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance. It delivers near Pro level reasoning and tool...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","file","audio","video"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.0000015","image":"0.00000025","audio":"0.0000005","web_search":"0.014","internal_reasoning":"0.0000015"},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3-flash-preview-20251217/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"agenticslides","elo":1073,"win_rate":39.3,"rank":8},{"arena":"agents","category":"agenticslides(python-pptx)","elo":1075,"win_rate":39.3,"rank":8},{"arena":"agents","category":"androidnative","elo":1066,"win_rate":48.1,"rank":32},{"arena":"agents","category":"fullstack","elo":1072,"win_rate":47.4,"rank":33},{"arena":"agents","category":"godotgamedev","elo":1161,"win_rate":50.6,"rank":13},{"arena":"agents","category":"mobileapps","elo":1117,"win_rate":46.6,"rank":36},{"arena":"agents","category":"python-pptxslides","elo":1036,"win_rate":38.3,"rank":27},{"arena":"agents","category":"webapps","elo":1124,"win_rate":48.5,"rank":31},{"arena":"models","category":"3d","elo":1212,"win_rate":62.7,"rank":45},{"arena":"models","category":"codecategories","elo":1204,"win_rate":57.6,"rank":51},{"arena":"models","category":"gamedev","elo":1189,"win_rate":58.3,"rank":57},{"arena":"models","category":"website","elo":1208,"win_rate":57,"rank":51}]},"reasoning":{"mandatory":false,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"nvidia/nemotron-3-nano-30b-a3b","canonical_slug":"nvidia/nemotron-3-nano-30b-a3b","hugging_face_id":"nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16","name":"NVIDIA: Nemotron 3 Nano 30B A3B","created":1765731275,"description":"NVIDIA Nemotron 3 Nano 30B A3B is a small language MoE model with highest compute efficiency and accuracy for developers to build specialized agentic AI systems. The model is fully...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000005","completion":"0.0000002","input_cache_read":"0.00000003"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/nvidia/nemotron-3-nano-30b-a3b/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":8.9,"coding_index":14.4,"agentic_index":1}},"reasoning":{"mandatory":false}},{"id":"openai/gpt-5.2-chat","canonical_slug":"openai/gpt-5.2-chat-20251211","hugging_face_id":"","name":"OpenAI: GPT-5.2 Chat","created":1765389783,"description":"GPT-5.2 Chat (AKA Instant) is the fast, lightweight member of the 5.2 family, optimized for low-latency chat while retaining strong general intelligence. It uses adaptive reasoning to selectively “think” on...","context_length":128000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000175","completion":"0.000014","web_search":"0.01","input_cache_read":"0.000000175"},"top_provider":{"context_length":128000,"max_completion_tokens":32000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_completion_tokens","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.2-chat-20251211/endpoints"}},{"id":"openai/gpt-5.2-pro","canonical_slug":"openai/gpt-5.2-pro-20251211","hugging_face_id":"","name":"OpenAI: GPT-5.2 Pro","created":1765389780,"description":"GPT-5.2 Pro is OpenAI’s most advanced model, offering major improvements in agentic coding and long context performance over GPT-5 Pro. It is optimized for complex tasks that require step-by-step reasoning,...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000021","completion":"0.000168","web_search":"0.01"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.2-pro-20251211/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium"],"default_effort":"medium"}},{"id":"openai/gpt-5.2-pro:batch","canonical_slug":"openai/gpt-5.2-pro-20251211","hugging_face_id":"","name":"OpenAI: GPT-5.2 Pro (batch)","created":1765389780,"description":"GPT-5.2 Pro is OpenAI’s most advanced model, offering major improvements in agentic coding and long context performance over GPT-5 Pro. It is optimized for complex tasks that require step-by-step reasoning,...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000105","completion":"0.000084","web_search":"0.01"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.2-pro-20251211/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium"],"default_effort":"medium"}},{"id":"openai/gpt-5.2","canonical_slug":"openai/gpt-5.2-20251211","hugging_face_id":"","name":"OpenAI: GPT-5.2","created":1765389775,"description":"GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1. It uses adaptive reasoning to allocate computation dynamically, responding quickly...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000175","completion":"0.000014","web_search":"0.01","input_cache_read":"0.000000175"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.2-20251211/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"website","elo":1207,"win_rate":54.4,"rank":52},{"arena":"agents","category":"androidnative","elo":1075,"win_rate":49.1,"rank":31},{"arena":"agents","category":"fullstack","elo":1051,"win_rate":44.3,"rank":37},{"arena":"agents","category":"godotgamedev","elo":1142,"win_rate":48.1,"rank":17},{"arena":"agents","category":"mobileapps","elo":1121,"win_rate":46.8,"rank":34},{"arena":"agents","category":"webapps","elo":1099,"win_rate":45.5,"rank":33},{"arena":"models","category":"3d","elo":1107,"win_rate":41.5,"rank":85},{"arena":"models","category":"codecategories","elo":1185,"win_rate":49.6,"rank":63},{"arena":"models","category":"dataviz","elo":1217,"win_rate":56.1,"rank":48},{"arena":"models","category":"gamedev","elo":1218,"win_rate":56,"rank":44},{"arena":"models","category":"uicomponent","elo":1206,"win_rate":51.3,"rank":51},{"arena":"models","category":"asciiart","elo":1170,"win_rate":50.7,"rank":33},{"arena":"models","category":"svg","elo":1164,"win_rate":53.7,"rank":44}]},"reasoning":{"mandatory":false,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"openai/gpt-5.2:batch","canonical_slug":"openai/gpt-5.2-20251211","hugging_face_id":"","name":"OpenAI: GPT-5.2 (batch)","created":1765389775,"description":"GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1. It uses adaptive reasoning to allocate computation dynamically, responding quickly...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000000875","completion":"0.000007","web_search":"0.01","input_cache_read":"0.0000000875"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.2-20251211/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"website","elo":1207,"win_rate":54.4,"rank":52},{"arena":"agents","category":"androidnative","elo":1075,"win_rate":49.1,"rank":31},{"arena":"agents","category":"fullstack","elo":1051,"win_rate":44.3,"rank":37},{"arena":"agents","category":"godotgamedev","elo":1142,"win_rate":48.1,"rank":17},{"arena":"agents","category":"mobileapps","elo":1121,"win_rate":46.8,"rank":34},{"arena":"agents","category":"webapps","elo":1099,"win_rate":45.5,"rank":33},{"arena":"models","category":"3d","elo":1107,"win_rate":41.5,"rank":85},{"arena":"models","category":"codecategories","elo":1185,"win_rate":49.6,"rank":63},{"arena":"models","category":"dataviz","elo":1217,"win_rate":56.1,"rank":48},{"arena":"models","category":"gamedev","elo":1218,"win_rate":56,"rank":44},{"arena":"models","category":"uicomponent","elo":1206,"win_rate":51.3,"rank":51},{"arena":"models","category":"asciiart","elo":1170,"win_rate":50.7,"rank":33},{"arena":"models","category":"svg","elo":1164,"win_rate":53.7,"rank":44}]},"reasoning":{"mandatory":false,"supported_efforts":["xhigh","high","medium","low","none"],"default_effort":"medium"}},{"id":"mistralai/devstral-2512","canonical_slug":"mistralai/devstral-2512","hugging_face_id":"mistralai/Devstral-2-123B-Instruct-2512","name":"Mistral: Devstral 2 2512","created":1765285419,"description":"Devstral 2 is a state-of-the-art open-source model by Mistral AI specializing in agentic coding. It is a 123B-parameter dense transformer model supporting a 256K context window. Devstral 2 supports exploring...","context_length":262144,"architecture":{"modality":"text+file->text","input_modalities":["text","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.0000004","completion":"0.000002","input_cache_read":"0.00000004"},"top_provider":{"context_length":262144,"max_completion_tokens":209715,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/devstral-2512/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":9.4,"coding_index":31.3,"agentic_index":4.9}}},{"id":"relace/relace-search","canonical_slug":"relace/relace-search-20251208","hugging_face_id":null,"name":"Relace: Relace Search","created":1765213560,"description":"The relace-search model uses 4-12 `view_file` and `grep` tools in parallel to explore a codebase and return relevant files to the user request. In contrast to RAG, relace-search performs agentic...","context_length":256000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000003"},"top_provider":{"context_length":256000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_tokens","response_format","seed","stop","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/relace/relace-search-20251208/endpoints"}},{"id":"z-ai/glm-4.6v","canonical_slug":"z-ai/glm-4.6-20251208","hugging_face_id":"zai-org/GLM-4.6V","name":"Z.ai: GLM 4.6V","created":1765207462,"description":"GLM-4.6V is a large multimodal model designed for high-fidelity visual understanding and long-context reasoning across images, documents, and mixed media. It supports up to 128K tokens, processes complex page layouts...","context_length":131072,"architecture":{"modality":"text+image+video->text","input_modalities":["image","text","video"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000009","input_cache_read":"0.000000055"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":0.8,"top_p":0.6,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-4.6-20251208/endpoints"},"reasoning":{"mandatory":false}},{"id":"openrouter/bodybuilder","canonical_slug":"openrouter/bodybuilder","hugging_face_id":"","name":"Body Builder (beta)","created":1764903653,"description":"Transform your natural language requests into structured OpenRouter API request objects. Describe what you want to accomplish with AI models, and Body Builder will construct the appropriate API calls. Example:...","context_length":128000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"-1","completion":"-1"},"top_provider":{"context_length":null,"max_completion_tokens":null,"is_moderated":false},"per_request_limits":null,"supported_parameters":[],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openrouter/bodybuilder/endpoints"}},{"id":"openai/gpt-5.1-codex-max","canonical_slug":"openai/gpt-5.1-codex-max-20251204","hugging_face_id":"","name":"OpenAI: GPT-5.1-Codex-Max","created":1764878934,"description":"GPT-5.1-Codex-Max is OpenAI’s latest agentic coding model, designed for long-running, high-context software development tasks. It is based on an updated version of the 5.1 reasoning stack and trained on agentic...","context_length":400000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.00001","web_search":"0.01","input_cache_read":"0.000000125"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.1-codex-max-20251204/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["xhigh","high","medium","low"],"default_effort":"medium"}},{"id":"amazon/nova-2-lite-v1","canonical_slug":"amazon/nova-2-lite-v1","hugging_face_id":"","name":"Amazon: Nova 2 Lite","created":1764696672,"description":"Nova 2 Lite is a fast, cost-effective reasoning model for everyday workloads that can process text, images, and videos to generate text. Nova 2 Lite demonstrates standout capabilities in processing...","context_length":1000000,"architecture":{"modality":"text+image+file+video->text","input_modalities":["text","image","video","file"],"output_modalities":["text"],"tokenizer":"Nova","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000025"},"top_provider":{"context_length":1000000,"max_completion_tokens":65535,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/amazon/nova-2-lite-v1/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":23,"agentic_index":null}},"reasoning":{"mandatory":false}},{"id":"mistralai/ministral-14b-2512","canonical_slug":"mistralai/ministral-14b-2512","hugging_face_id":"mistralai/Ministral-3-14B-Instruct-2512","name":"Mistral: Ministral 3 14B 2512","created":1764681735,"description":"The largest model in the Ministral 3 family, Ministral 3 14B offers frontier capabilities and performance comparable to its larger Mistral Small 3.2 24B counterpart. A powerful and efficient language...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.0000002","input_cache_read":"0.00000002"},"top_provider":{"context_length":262144,"max_completion_tokens":209715,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/ministral-14b-2512/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1018,"win_rate":39.6,"rank":102},{"arena":"models","category":"codecategories","elo":1080,"win_rate":44,"rank":98},{"arena":"models","category":"gamedev","elo":1061,"win_rate":43.6,"rank":98},{"arena":"models","category":"website","elo":1093,"win_rate":44.8,"rank":99}],"artificial_analysis":{"intelligence_index":6,"coding_index":14.4,"agentic_index":1.1}}},{"id":"mistralai/ministral-8b-2512","canonical_slug":"mistralai/ministral-8b-2512","hugging_face_id":"mistralai/Ministral-3-8B-Instruct-2512","name":"Mistral: Ministral 3 8B 2512","created":1764681654,"description":"A balanced model in the Ministral 3 family, Ministral 3 8B is a powerful, efficient tiny language model with vision capabilities.","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.00000015","input_cache_read":"0.000000015"},"top_provider":{"context_length":262144,"max_completion_tokens":209715,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/ministral-8b-2512/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1060,"win_rate":46.2,"rank":96},{"arena":"models","category":"codecategories","elo":1068,"win_rate":42.9,"rank":99},{"arena":"models","category":"gamedev","elo":1013,"win_rate":38.7,"rank":108},{"arena":"models","category":"website","elo":1077,"win_rate":42.9,"rank":101}],"artificial_analysis":{"intelligence_index":5.5,"coding_index":9.7,"agentic_index":0.6}}},{"id":"mistralai/ministral-8b-2512:batch","canonical_slug":"mistralai/ministral-8b-2512","hugging_face_id":"mistralai/Ministral-3-8B-Instruct-2512","name":"Mistral: Ministral 3 8B 2512 (batch)","created":1764681654,"description":"A balanced model in the Ministral 3 family, Ministral 3 8B is a powerful, efficient tiny language model with vision capabilities.","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.000000075","completion":"0.000000075","input_cache_read":"0.0000000075"},"top_provider":{"context_length":262144,"max_completion_tokens":209715,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/ministral-8b-2512/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1060,"win_rate":46.2,"rank":96},{"arena":"models","category":"codecategories","elo":1068,"win_rate":42.9,"rank":99},{"arena":"models","category":"gamedev","elo":1013,"win_rate":38.7,"rank":108},{"arena":"models","category":"website","elo":1077,"win_rate":42.9,"rank":101}],"artificial_analysis":{"intelligence_index":5.5,"coding_index":9.7,"agentic_index":0.6}}},{"id":"mistralai/ministral-3b-2512","canonical_slug":"mistralai/ministral-3b-2512","hugging_face_id":"mistralai/Ministral-3-3B-Instruct-2512","name":"Mistral: Ministral 3 3B 2512","created":1764681560,"description":"The smallest model in the Ministral 3 family, Ministral 3 3B is a powerful, efficient tiny language model with vision capabilities.","context_length":131072,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000001","input_cache_read":"0.00000001"},"top_provider":{"context_length":131072,"max_completion_tokens":104857,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/ministral-3b-2512/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":991,"win_rate":35.9,"rank":107},{"arena":"models","category":"codecategories","elo":1025,"win_rate":37.3,"rank":109},{"arena":"models","category":"gamedev","elo":971,"win_rate":33,"rank":115},{"arena":"models","category":"website","elo":1040,"win_rate":38.2,"rank":110}],"artificial_analysis":{"intelligence_index":4.8,"coding_index":4.8,"agentic_index":0.8}}},{"id":"mistralai/mistral-large-2512","canonical_slug":"mistralai/mistral-large-2512","hugging_face_id":"","name":"Mistral: Mistral Large 3 2512","created":1764624472,"description":"Mistral Large 3 2512 is Mistral’s most capable model to date, featuring a sparse mixture-of-experts architecture with 41B active parameters (675B total), and released under the Apache 2.0 license.","context_length":262144,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.0000005","completion":"0.0000015","input_cache_read":"0.00000005"},"top_provider":{"context_length":262144,"max_completion_tokens":209715,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.0645,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-large-2512/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1128,"win_rate":46.9,"rank":76},{"arena":"models","category":"asciiart","elo":1087,"win_rate":40.3,"rank":61},{"arena":"models","category":"codecategories","elo":1157,"win_rate":47.6,"rank":74},{"arena":"models","category":"dataviz","elo":1152,"win_rate":45.7,"rank":75},{"arena":"models","category":"gamedev","elo":1103,"win_rate":41.5,"rank":87},{"arena":"models","category":"svg","elo":1017,"win_rate":38,"rank":79},{"arena":"models","category":"uicomponent","elo":1117,"win_rate":43,"rank":81},{"arena":"models","category":"website","elo":1174,"win_rate":49.5,"rank":73}],"artificial_analysis":{"intelligence_index":9.7,"coding_index":20.1,"agentic_index":2.4}}},{"id":"mistralai/mistral-large-2512:batch","canonical_slug":"mistralai/mistral-large-2512","hugging_face_id":"","name":"Mistral: Mistral Large 3 2512 (batch)","created":1764624472,"description":"Mistral Large 3 2512 is Mistral’s most capable model to date, featuring a sparse mixture-of-experts architecture with 41B active parameters (675B total), and released under the Apache 2.0 license.","context_length":262144,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.00000075","input_cache_read":"0.000000025"},"top_provider":{"context_length":262144,"max_completion_tokens":209715,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.0645,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-large-2512/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1128,"win_rate":46.9,"rank":76},{"arena":"models","category":"asciiart","elo":1087,"win_rate":40.3,"rank":61},{"arena":"models","category":"codecategories","elo":1157,"win_rate":47.6,"rank":74},{"arena":"models","category":"dataviz","elo":1152,"win_rate":45.7,"rank":75},{"arena":"models","category":"gamedev","elo":1103,"win_rate":41.5,"rank":87},{"arena":"models","category":"svg","elo":1017,"win_rate":38,"rank":79},{"arena":"models","category":"uicomponent","elo":1117,"win_rate":43,"rank":81},{"arena":"models","category":"website","elo":1174,"win_rate":49.5,"rank":73}],"artificial_analysis":{"intelligence_index":9.7,"coding_index":20.1,"agentic_index":2.4}}},{"id":"deepseek/deepseek-v3.2","canonical_slug":"deepseek/deepseek-v3.2-20251201","hugging_face_id":"deepseek-ai/DeepSeek-V3.2","name":"DeepSeek: DeepSeek V3.2","created":1764594642,"description":"DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use performance. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...","context_length":163840,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.000000269","completion":"0.0000004","input_cache_read":"0.0000001345"},"top_provider":{"context_length":163840,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v3.2-20251201/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1161,"win_rate":49.4,"rank":63},{"arena":"models","category":"asciiart","elo":1100,"win_rate":40.5,"rank":59},{"arena":"models","category":"codecategories","elo":1179,"win_rate":49.3,"rank":68},{"arena":"models","category":"dataviz","elo":1175,"win_rate":48.2,"rank":67},{"arena":"models","category":"gamedev","elo":1154,"win_rate":46.5,"rank":71},{"arena":"models","category":"svg","elo":1057,"win_rate":40.8,"rank":70},{"arena":"models","category":"uicomponent","elo":1164,"win_rate":46.8,"rank":67},{"arena":"models","category":"website","elo":1187,"win_rate":50.2,"rank":66}],"artificial_analysis":{"intelligence_index":null,"coding_index":44.2,"agentic_index":null}},"reasoning":{"mandatory":false,"default_enabled":false}},{"id":"anthropic/claude-opus-4.5","canonical_slug":"anthropic/claude-4.5-opus-20251124","hugging_face_id":"","name":"Anthropic: Claude Opus 4.5","created":1764010580,"description":"Claude Opus 4.5 is Anthropic’s frontier reasoning model optimized for complex software engineering, agentic workflows, and long-horizon computer use. It offers strong multimodal capabilities, competitive performance across real-world coding and...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000025","web_search":"0.01","input_cache_read":"0.0000005","input_cache_write":"0.00000625","input_cache_write_1h":"0.00001"},"top_provider":{"context_length":200000,"max_completion_tokens":64000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.5-opus-20251124/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1242,"win_rate":58.5,"rank":33},{"arena":"models","category":"asciiart","elo":1208,"win_rate":54.8,"rank":24},{"arena":"models","category":"codecategories","elo":1257,"win_rate":59.5,"rank":34},{"arena":"models","category":"dataviz","elo":1259,"win_rate":58.5,"rank":28},{"arena":"models","category":"gamedev","elo":1252,"win_rate":59.3,"rank":32},{"arena":"models","category":"svg","elo":1205,"win_rate":58.7,"rank":30},{"arena":"models","category":"uicomponent","elo":1253,"win_rate":58.1,"rank":34},{"arena":"models","category":"website","elo":1259,"win_rate":59.7,"rank":35},{"arena":"agents","category":"androidnative","elo":1192,"win_rate":65.5,"rank":15},{"arena":"agents","category":"fullstack","elo":1165,"win_rate":60.2,"rank":24},{"arena":"agents","category":"mobileapps","elo":1216,"win_rate":60.7,"rank":15},{"arena":"agents","category":"webapps","elo":1173,"win_rate":55.1,"rank":24}]},"reasoning":{"mandatory":false}},{"id":"anthropic/claude-opus-4.5:batch","canonical_slug":"anthropic/claude-4.5-opus-20251124","hugging_face_id":"","name":"Anthropic: Claude Opus 4.5 (batch)","created":1764010580,"description":"Claude Opus 4.5 is Anthropic’s frontier reasoning model optimized for complex software engineering, agentic workflows, and long-horizon computer use. It offers strong multimodal capabilities, competitive performance across real-world coding and...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["file","image","text"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.0000125","web_search":"0.01","input_cache_read":"0.00000025","input_cache_write":"0.000003125","input_cache_write_1h":"0.000005"},"top_provider":{"context_length":200000,"max_completion_tokens":64000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","stop","structured_outputs","temperature","tool_choice","tools","verbosity"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.5-opus-20251124/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1242,"win_rate":58.5,"rank":33},{"arena":"models","category":"asciiart","elo":1208,"win_rate":54.8,"rank":24},{"arena":"models","category":"codecategories","elo":1257,"win_rate":59.5,"rank":34},{"arena":"models","category":"dataviz","elo":1259,"win_rate":58.5,"rank":28},{"arena":"models","category":"gamedev","elo":1252,"win_rate":59.3,"rank":32},{"arena":"models","category":"svg","elo":1205,"win_rate":58.7,"rank":30},{"arena":"models","category":"uicomponent","elo":1253,"win_rate":58.1,"rank":34},{"arena":"models","category":"website","elo":1259,"win_rate":59.7,"rank":35},{"arena":"agents","category":"androidnative","elo":1192,"win_rate":65.5,"rank":15},{"arena":"agents","category":"fullstack","elo":1165,"win_rate":60.2,"rank":24},{"arena":"agents","category":"mobileapps","elo":1216,"win_rate":60.7,"rank":15},{"arena":"agents","category":"webapps","elo":1173,"win_rate":55.1,"rank":24}]},"reasoning":{"mandatory":false}},{"id":"google/gemini-3-pro-image-preview","canonical_slug":"google/gemini-3-pro-image-preview-20251120","hugging_face_id":"","name":"Google: Nano Banana Pro (Gemini 3 Pro Image Preview)","created":1763653797,"description":"Nano Banana Pro is Google’s most advanced image-generation and editing model, built on Gemini 3 Pro. It extends the original Nano Banana with significantly improved multimodal reasoning, real-world grounding, and...","context_length":65536,"architecture":{"modality":"text+image->text+image","input_modalities":["image","text"],"output_modalities":["image","text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000012","image":"0.000002","image_output":"0.00012","audio":"0.000002","input_audio_cache":"0.0000002","web_search":"0.014","internal_reasoning":"0.000012","input_cache_read":"0.0000002","input_cache_write":"0.000000375"},"top_provider":{"context_length":65536,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","stop","structured_outputs","temperature","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-3-pro-image-preview-20251120/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"graphicdesign","elo":1250,"win_rate":65.8,"rank":5},{"arena":"models","category":"image","elo":1247,"win_rate":62.1,"rank":5},{"arena":"models","category":"logo","elo":1240,"win_rate":61,"rank":6},{"arena":"models","category":"imageediting","elo":1254,"win_rate":66,"rank":3}]},"reasoning":{"mandatory":true}},{"id":"openai/gpt-5.1","canonical_slug":"openai/gpt-5.1-20251113","hugging_face_id":"","name":"OpenAI: GPT-5.1","created":1763060305,"description":"GPT-5.1 is the latest frontier-grade model in the GPT-5 series, offering stronger general-purpose reasoning, improved instruction adherence, and a more natural conversational style compared to GPT-5. It uses adaptive reasoning...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.00001","web_search":"0.01","input_cache_read":"0.000000125"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.1-20251113/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1089,"win_rate":43.7,"rank":90},{"arena":"models","category":"asciiart","elo":1136,"win_rate":48.6,"rank":50},{"arena":"models","category":"codecategories","elo":1186,"win_rate":53,"rank":62},{"arena":"models","category":"dataviz","elo":1220,"win_rate":58,"rank":46},{"arena":"models","category":"gamedev","elo":1201,"win_rate":55.9,"rank":49},{"arena":"models","category":"svg","elo":1174,"win_rate":57.4,"rank":41},{"arena":"models","category":"uicomponent","elo":1182,"win_rate":52.9,"rank":60},{"arena":"models","category":"website","elo":1199,"win_rate":54,"rank":60}],"artificial_analysis":{"intelligence_index":null,"coding_index":49.4,"agentic_index":null}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","medium","low","none"],"default_effort":"none"}},{"id":"openai/gpt-5.1:batch","canonical_slug":"openai/gpt-5.1-20251113","hugging_face_id":"","name":"OpenAI: GPT-5.1 (batch)","created":1763060305,"description":"GPT-5.1 is the latest frontier-grade model in the GPT-5 series, offering stronger general-purpose reasoning, improved instruction adherence, and a more natural conversational style compared to GPT-5. It uses adaptive reasoning...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000000625","completion":"0.000005","web_search":"0.01","input_cache_read":"0.0000000625"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.1-20251113/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1089,"win_rate":43.7,"rank":90},{"arena":"models","category":"asciiart","elo":1136,"win_rate":48.6,"rank":50},{"arena":"models","category":"codecategories","elo":1186,"win_rate":53,"rank":62},{"arena":"models","category":"dataviz","elo":1220,"win_rate":58,"rank":46},{"arena":"models","category":"gamedev","elo":1201,"win_rate":55.9,"rank":49},{"arena":"models","category":"svg","elo":1174,"win_rate":57.4,"rank":41},{"arena":"models","category":"uicomponent","elo":1182,"win_rate":52.9,"rank":60},{"arena":"models","category":"website","elo":1199,"win_rate":54,"rank":60}],"artificial_analysis":{"intelligence_index":null,"coding_index":49.4,"agentic_index":null}},"reasoning":{"mandatory":false,"default_enabled":true,"supported_efforts":["high","medium","low","none"],"default_effort":"none"}},{"id":"openai/gpt-5.1-codex","canonical_slug":"openai/gpt-5.1-codex-20251113","hugging_face_id":"","name":"OpenAI: GPT-5.1-Codex","created":1763060298,"description":"GPT-5.1-Codex is a specialized version of GPT-5.1 optimized for software engineering and coding workflows. It is designed for both interactive development sessions and long, independent execution of complex engineering tasks....","context_length":400000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.00001","web_search":"0.01","input_cache_read":"0.00000013"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.1-codex-20251113/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"fullstack","elo":1086,"win_rate":44.5,"rank":30},{"arena":"agents","category":"mobileapps","elo":1191,"win_rate":53.4,"rank":19},{"arena":"agents","category":"webapps","elo":1032,"win_rate":44,"rank":38},{"arena":"models","category":"codecategories","elo":1167,"win_rate":54.6,"rank":71},{"arena":"models","category":"dataviz","elo":1229,"win_rate":55.9,"rank":44},{"arena":"models","category":"gamedev","elo":1161,"win_rate":50.5,"rank":66},{"arena":"models","category":"uicomponent","elo":1252,"win_rate":60,"rank":36},{"arena":"models","category":"website","elo":1174,"win_rate":56.1,"rank":72}]},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"openai/gpt-5.1-codex-mini","canonical_slug":"openai/gpt-5.1-codex-mini-20251113","hugging_face_id":"","name":"OpenAI: GPT-5.1-Codex-Mini","created":1763057820,"description":"GPT-5.1-Codex-Mini is a smaller and faster version of GPT-5.1-Codex","context_length":400000,"architecture":{"modality":"text+image->text","input_modalities":["image","text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.000002","web_search":"0.01","input_cache_read":"0.00000003"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5.1-codex-mini-20251113/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1017,"win_rate":32.8,"rank":103},{"arena":"models","category":"asciiart","elo":1123,"win_rate":43,"rank":54},{"arena":"models","category":"codecategories","elo":1109,"win_rate":41.5,"rank":92},{"arena":"models","category":"dataviz","elo":1113,"win_rate":40.6,"rank":90},{"arena":"models","category":"gamedev","elo":1115,"win_rate":43.4,"rank":85},{"arena":"models","category":"svg","elo":1003,"win_rate":35.3,"rank":82},{"arena":"models","category":"uicomponent","elo":1096,"win_rate":41,"rank":88},{"arena":"models","category":"website","elo":1123,"win_rate":42.8,"rank":92}]},"reasoning":{"mandatory":false,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"moonshotai/kimi-k2-thinking","canonical_slug":"moonshotai/kimi-k2-thinking-20251106","hugging_face_id":"moonshotai/Kimi-K2-Thinking","name":"MoonshotAI: Kimi K2 Thinking","created":1762440622,"description":"Kimi K2 Thinking is Moonshot AI’s most advanced open reasoning model to date, extending the K2 series into agentic, long-horizon reasoning. Built on the trillion-parameter Mixture-of-Experts (MoE) architecture introduced in...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000006","completion":"0.0000025","input_cache_read":"0.00000015"},"top_provider":{"context_length":262144,"max_completion_tokens":98304,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/moonshotai/kimi-k2-thinking-20251106/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"website","elo":1125,"win_rate":48.8,"rank":91}],"artificial_analysis":{"intelligence_index":null,"coding_index":21,"agentic_index":null}},"reasoning":{"mandatory":true,"default_enabled":true}},{"id":"amazon/nova-premier-v1","canonical_slug":"amazon/nova-premier-v1","hugging_face_id":"","name":"Amazon: Nova Premier 1.0","created":1761950332,"description":"Amazon Nova Premier is the most capable of Amazon’s multimodal models for complex reasoning tasks and for use as the best teacher for distilling custom models.","context_length":1000000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Nova","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.0000125","input_cache_read":"0.000000625"},"top_provider":{"context_length":1000000,"max_completion_tokens":32000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_tokens","stop","temperature","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/amazon/nova-premier-v1/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"website","elo":847,"win_rate":26.2,"rank":127}]}},{"id":"perplexity/sonar-pro-search","canonical_slug":"perplexity/sonar-pro-search","hugging_face_id":"","name":"Perplexity: Sonar Pro Search","created":1761854366,"description":"Exclusively available on the OpenRouter API, Sonar Pro's new Pro Search mode is Perplexity's most advanced agentic search system. It is designed for deeper reasoning and analysis. Pricing is based...","context_length":200000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000003","completion":"0.000015","web_search":"0.018"},"top_provider":{"context_length":200000,"max_completion_tokens":8000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","structured_outputs","temperature","top_k","top_p","web_search_options"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/perplexity/sonar-pro-search/endpoints"},"reasoning":{"mandatory":true}},{"id":"mistralai/voxtral-small-24b-2507","canonical_slug":"mistralai/voxtral-small-24b-2507","hugging_face_id":"mistralai/Voxtral-Small-24B-2507","name":"Mistral: Voxtral Small 24B 2507","created":1761835144,"description":"Voxtral Small is an enhancement of Mistral Small 3, incorporating state-of-the-art audio input capabilities while retaining best-in-class text performance. It excels at speech transcription, translation and audio understanding. Input audio...","context_length":32768,"architecture":{"modality":"text+file+audio->text","input_modalities":["text","audio","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000003","audio":"0.0001","input_cache_read":"0.00000001"},"top_provider":{"context_length":32768,"max_completion_tokens":26214,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.2,"top_p":0.95,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/mistralai/voxtral-small-24b-2507/endpoints"}},{"id":"openai/gpt-oss-safeguard-20b","canonical_slug":"openai/gpt-oss-safeguard-20b","hugging_face_id":"openai/gpt-oss-safeguard-20b","name":"OpenAI: gpt-oss-safeguard-20b","created":1761752836,"description":"gpt-oss-safeguard-20b is a safety reasoning model from OpenAI built upon gpt-oss-20b. This open-weight, 21B-parameter Mixture-of-Experts (MoE) model offers lower latency for safety tasks like content classification, LLM filtering, and trust...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000000075","completion":"0.0000003","input_cache_read":"0.0000000375"},"top_provider":{"context_length":131072,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-oss-safeguard-20b/endpoints"},"reasoning":{"mandatory":true}},{"id":"minimax/minimax-m2","canonical_slug":"minimax/minimax-m2","hugging_face_id":"MiniMaxAI/MiniMax-M2","name":"MiniMax: MiniMax M2","created":1761252093,"description":"MiniMax-M2 is a compact, high-efficiency large language model optimized for end-to-end coding and agentic workflows. With 10 billion activated parameters (230 billion total), it delivers near-frontier intelligence across general reasoning,...","context_length":204800,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000255","completion":"0.00000102"},"top_provider":{"context_length":204800,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/minimax/minimax-m2/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1126,"win_rate":48.3,"rank":78},{"arena":"models","category":"codecategories","elo":1150,"win_rate":48.1,"rank":81},{"arena":"models","category":"dataviz","elo":1156,"win_rate":50,"rank":74},{"arena":"models","category":"gamedev","elo":1141,"win_rate":48.1,"rank":75},{"arena":"models","category":"svg","elo":1125,"win_rate":55.3,"rank":55},{"arena":"models","category":"uicomponent","elo":1152,"win_rate":49.2,"rank":72},{"arena":"models","category":"website","elo":1155,"win_rate":48,"rank":80}]},"reasoning":{"mandatory":true}},{"id":"qwen/qwen3-vl-32b-instruct","canonical_slug":"qwen/qwen3-vl-32b-instruct","hugging_face_id":"Qwen/Qwen3-VL-32B-Instruct","name":"Qwen: Qwen3 VL 32B Instruct","created":1761231332,"description":"Qwen3-VL-32B-Instruct is a large-scale multimodal vision-language model designed for high-precision understanding and reasoning across text, images, and video. With 32 billion parameters, it combines deep visual perception with advanced text...","context_length":131072,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.000000104","completion":"0.000000416"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.7,"top_p":0.8,"top_k":20,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":1},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-vl-32b-instruct/endpoints"}},{"id":"ibm-granite/granite-4.0-h-micro","canonical_slug":"ibm-granite/granite-4.0-h-micro","hugging_face_id":"ibm-granite/granite-4.0-h-micro","name":"IBM: Granite 4.0 Micro","created":1760927695,"description":"Granite-4.0-H-Micro is a 3B parameter from the Granite 4 family of models. These models are the latest in a series of models released by IBM. They are fine-tuned for long...","context_length":131000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000000017","completion":"0.000000112"},"top_provider":{"context_length":131000,"max_completion_tokens":117900,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","temperature","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/ibm-granite/granite-4.0-h-micro/endpoints"}},{"id":"openai/gpt-5-image-mini","canonical_slug":"openai/gpt-5-image-mini","hugging_face_id":"","name":"OpenAI: GPT-5 Image Mini","created":1760624583,"description":"GPT-5 Image Mini combines OpenAI's advanced language capabilities, powered by [GPT-5 Mini](https://openrouter.ai/openai/gpt-5-mini), with GPT Image 1 Mini for efficient image generation. This natively multimodal model features superior instruction following, text...","context_length":400000,"architecture":{"modality":"text+image+file->text+image","input_modalities":["file","image","text"],"output_modalities":["image","text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.000002","image_output":"0.000008","web_search":"0.01","input_cache_read":"0.00000025"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5-image-mini/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"graphicdesign","elo":1164,"win_rate":47.7,"rank":11},{"arena":"models","category":"image","elo":1186,"win_rate":51,"rank":11},{"arena":"models","category":"logo","elo":1209,"win_rate":51.4,"rank":8}]},"reasoning":{"mandatory":true}},{"id":"anthropic/claude-haiku-4.5","canonical_slug":"anthropic/claude-4.5-haiku-20251001","hugging_face_id":"","name":"Anthropic: Claude Haiku 4.5","created":1760547638,"description":"Claude Haiku 4.5 is Anthropic’s fastest and most efficient model, delivering near-frontier intelligence at a fraction of the cost and latency of larger Claude models. Matching Claude Sonnet 4’s performance...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000005","web_search":"0.01","input_cache_read":"0.0000001","input_cache_write":"0.00000125","input_cache_write_1h":"0.000002"},"top_provider":{"context_length":200000,"max_completion_tokens":64000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.5-haiku-20251001/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1101,"win_rate":41.1,"rank":88},{"arena":"models","category":"asciiart","elo":1158,"win_rate":49.3,"rank":41},{"arena":"models","category":"codecategories","elo":1130,"win_rate":44.9,"rank":85},{"arena":"models","category":"dataviz","elo":1140,"win_rate":45.7,"rank":82},{"arena":"models","category":"gamedev","elo":1120,"win_rate":44.7,"rank":81},{"arena":"models","category":"svg","elo":1050,"win_rate":39.1,"rank":72},{"arena":"models","category":"uicomponent","elo":1115,"win_rate":42.6,"rank":82},{"arena":"models","category":"website","elo":1134,"win_rate":45.1,"rank":87}],"artificial_analysis":{"intelligence_index":17.6,"coding_index":43.9,"agentic_index":10.3}},"reasoning":{"mandatory":false}},{"id":"anthropic/claude-haiku-4.5:batch","canonical_slug":"anthropic/claude-4.5-haiku-20251001","hugging_face_id":"","name":"Anthropic: Claude Haiku 4.5 (batch)","created":1760547638,"description":"Claude Haiku 4.5 is Anthropic’s fastest and most efficient model, delivering near-frontier intelligence at a fraction of the cost and latency of larger Claude models. Matching Claude Sonnet 4’s performance...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.0000005","completion":"0.0000025","web_search":"0.01","input_cache_read":"0.00000005","input_cache_write":"0.000000625","input_cache_write_1h":"0.000001"},"top_provider":{"context_length":200000,"max_completion_tokens":64000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.5-haiku-20251001/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1101,"win_rate":41.1,"rank":88},{"arena":"models","category":"asciiart","elo":1158,"win_rate":49.3,"rank":41},{"arena":"models","category":"codecategories","elo":1130,"win_rate":44.9,"rank":85},{"arena":"models","category":"dataviz","elo":1140,"win_rate":45.7,"rank":82},{"arena":"models","category":"gamedev","elo":1120,"win_rate":44.7,"rank":81},{"arena":"models","category":"svg","elo":1050,"win_rate":39.1,"rank":72},{"arena":"models","category":"uicomponent","elo":1115,"win_rate":42.6,"rank":82},{"arena":"models","category":"website","elo":1134,"win_rate":45.1,"rank":87}],"artificial_analysis":{"intelligence_index":17.6,"coding_index":43.9,"agentic_index":10.3}},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3-vl-8b-thinking","canonical_slug":"qwen/qwen3-vl-8b-thinking","hugging_face_id":"Qwen/Qwen3-VL-8B-Thinking","name":"Qwen: Qwen3 VL 8B Thinking","created":1760463746,"description":"Qwen3-VL-8B-Thinking is the reasoning-optimized variant of the Qwen3-VL-8B multimodal model, designed for advanced visual and textual reasoning across complex scenes, documents, and temporal sequences. It integrates enhanced multimodal alignment and...","context_length":131072,"architecture":{"modality":"text+image->text","input_modalities":["image","text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000018","completion":"0.0000021"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":0.95},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-vl-8b-thinking/endpoints"},"reasoning":{"mandatory":true}},{"id":"qwen/qwen3-vl-8b-instruct","canonical_slug":"qwen/qwen3-vl-8b-instruct","hugging_face_id":"Qwen/Qwen3-VL-8B-Instruct","name":"Qwen: Qwen3 VL 8B Instruct","created":1760463308,"description":"Qwen3-VL-8B-Instruct is a multimodal vision-language model from the Qwen3-VL series, built for high-fidelity understanding and reasoning across text, images, and video. It features improved multimodal fusion with Interleaved-MRoPE for long-horizon...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["image","text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.000000117","completion":"0.000000455"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.7,"top_p":0.8,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-vl-8b-instruct/endpoints"}},{"id":"openai/gpt-5-image","canonical_slug":"openai/gpt-5-image","hugging_face_id":"","name":"OpenAI: GPT-5 Image","created":1760447986,"description":"[GPT-5](https://openrouter.ai/openai/gpt-5) Image combines OpenAI's GPT-5 model with state-of-the-art image generation capabilities. It offers major improvements in reasoning, code quality, and user experience while incorporating GPT Image 1's superior instruction following,...","context_length":400000,"architecture":{"modality":"text+image+file->text+image","input_modalities":["image","text","file"],"output_modalities":["image","text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00001","completion":"0.00001","image_output":"0.00004","web_search":"0.01","input_cache_read":"0.00000125"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5-image/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"graphicdesign","elo":1171,"win_rate":48.9,"rank":9},{"arena":"models","category":"image","elo":1194,"win_rate":53.9,"rank":9},{"arena":"models","category":"logo","elo":1202,"win_rate":52.7,"rank":9}]},"reasoning":{"mandatory":true}},{"id":"google/gemini-2.5-flash-image","canonical_slug":"google/gemini-2.5-flash-image","hugging_face_id":"","name":"Google: Nano Banana (Gemini 2.5 Flash Image)","created":1759870431,"description":"Gemini 2.5 Flash Image, a.k.a. \"Nano Banana,\" is now generally available. It is a state of the art image generation model with contextual understanding. It is capable of image generation,...","context_length":32768,"architecture":{"modality":"text+image->text+image","input_modalities":["image","text"],"output_modalities":["image","text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000025","image":"0.0000003","image_output":"0.00003","audio":"0.000001","input_audio_cache":"0.0000001","web_search":"0.014","internal_reasoning":"0.0000025","input_cache_read":"0.00000003","input_cache_write":"0.0000000833333333333333"},"top_provider":{"context_length":32768,"max_completion_tokens":8192,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_tokens","response_format","seed","stop","structured_outputs","temperature","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-2.5-flash-image/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"graphicdesign","elo":1167,"win_rate":56.9,"rank":10},{"arena":"models","category":"image","elo":1189,"win_rate":55.6,"rank":10},{"arena":"models","category":"logo","elo":1169,"win_rate":51.4,"rank":11}]}},{"id":"qwen/qwen3-vl-30b-a3b-thinking","canonical_slug":"qwen/qwen3-vl-30b-a3b-thinking","hugging_face_id":"Qwen/Qwen3-VL-30B-A3B-Thinking","name":"Qwen: Qwen3 VL 30B A3B Thinking","created":1759794479,"description":"Qwen3-VL-30B-A3B-Thinking is a multimodal model that unifies strong text generation with visual understanding for images and videos. Its Thinking variant enhances reasoning in STEM, math, and complex tasks. It excels...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.0000024"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.8,"top_p":0.95,"top_k":20,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":1},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-vl-30b-a3b-thinking/endpoints"},"reasoning":{"mandatory":true}},{"id":"qwen/qwen3-vl-30b-a3b-instruct","canonical_slug":"qwen/qwen3-vl-30b-a3b-instruct","hugging_face_id":"Qwen/Qwen3-VL-30B-A3B-Instruct","name":"Qwen: Qwen3 VL 30B A3B Instruct","created":1759794476,"description":"Qwen3-VL-30B-A3B-Instruct is a multimodal model that unifies strong text generation with visual understanding for images and videos. Its Instruct variant optimizes instruction-following for general multimodal tasks. It excels in perception...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000006"},"top_provider":{"context_length":262144,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.7,"top_p":0.8,"top_k":20,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":1},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-vl-30b-a3b-instruct/endpoints"}},{"id":"openai/gpt-5-pro","canonical_slug":"openai/gpt-5-pro-2025-10-06","hugging_face_id":"","name":"OpenAI: GPT-5 Pro","created":1759776663,"description":"GPT-5 Pro is OpenAI’s most advanced model, offering major improvements in reasoning, code quality, and user experience. It is optimized for complex tasks that require step-by-step reasoning, instruction following, and...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000015","completion":"0.00012","web_search":"0.01"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-09-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5-pro-2025-10-06/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["high"],"default_effort":"high"}},{"id":"openai/gpt-5-pro:batch","canonical_slug":"openai/gpt-5-pro-2025-10-06","hugging_face_id":"","name":"OpenAI: GPT-5 Pro (batch)","created":1759776663,"description":"GPT-5 Pro is OpenAI’s most advanced model, offering major improvements in reasoning, code quality, and user experience. It is optimized for complex tasks that require step-by-step reasoning, instruction following, and...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000075","completion":"0.00006","web_search":"0.01"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-09-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5-pro-2025-10-06/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["high"],"default_effort":"high"}},{"id":"z-ai/glm-4.6","canonical_slug":"z-ai/glm-4.6","hugging_face_id":"zai-org/GLM-4.6","name":"Z.ai: GLM 4.6","created":1759235576,"description":"Compared with GLM-4.5, this generation brings several key improvements: Longer context window: The context window has been expanded from 128K to 200K tokens, enabling the model to handle more complex...","context_length":204800,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000043","completion":"0.00000175","input_cache_read":"0.00000008"},"top_provider":{"context_length":198000,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":0.6,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-4.6/endpoints"},"benchmarks":{"design_arena":[{"arena":"agents","category":"androidnative","elo":1100,"win_rate":52.6,"rank":29},{"arena":"agents","category":"fullstack","elo":1038,"win_rate":42.3,"rank":38},{"arena":"agents","category":"godotgamedev","elo":1180,"win_rate":53.2,"rank":12},{"arena":"agents","category":"mobileapps","elo":1123,"win_rate":47.2,"rank":33},{"arena":"models","category":"3d","elo":1157,"win_rate":54.2,"rank":64},{"arena":"models","category":"codecategories","elo":1180,"win_rate":54.3,"rank":67},{"arena":"models","category":"dataviz","elo":1177,"win_rate":52.6,"rank":66},{"arena":"models","category":"gamedev","elo":1174,"win_rate":54.8,"rank":61},{"arena":"models","category":"svg","elo":1133,"win_rate":52.1,"rank":53},{"arena":"models","category":"uicomponent","elo":1163,"win_rate":52.5,"rank":68},{"arena":"models","category":"website","elo":1186,"win_rate":54.3,"rank":67}],"artificial_analysis":{"intelligence_index":null,"coding_index":45.8,"agentic_index":null}},"reasoning":{"mandatory":false}},{"id":"anthropic/claude-sonnet-4.5","canonical_slug":"anthropic/claude-4.5-sonnet-20250929","hugging_face_id":"","name":"Anthropic: Claude Sonnet 4.5","created":1759161676,"description":"Claude Sonnet 4.5 is Anthropic’s most advanced Sonnet model to date, optimized for real-world agents and coding workflows. It delivers state-of-the-art performance on coding benchmarks such as SWE-bench Verified, with...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000003","completion":"0.000015","web_search":"0.01","input_cache_read":"0.0000003","input_cache_write":"0.00000375","input_cache_write_1h":"0.000006","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000006","completion":"0.0000225","input_cache_read":"0.0000006","input_cache_write":"0.0000075","input_cache_write_1h":"0.000012"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":64000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":1,"top_p":1,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.5-sonnet-20250929/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1186,"win_rate":51,"rank":54},{"arena":"models","category":"asciiart","elo":1223,"win_rate":56.1,"rank":20},{"arena":"models","category":"codecategories","elo":1198,"win_rate":51.5,"rank":53},{"arena":"models","category":"dataviz","elo":1183,"win_rate":47.4,"rank":62},{"arena":"models","category":"gamedev","elo":1189,"win_rate":51.1,"rank":56},{"arena":"models","category":"svg","elo":1140,"win_rate":52.2,"rank":51},{"arena":"models","category":"uicomponent","elo":1188,"win_rate":49.4,"rank":57},{"arena":"models","category":"website","elo":1202,"win_rate":51.8,"rank":57},{"arena":"agents","category":"fullstack","elo":1059,"win_rate":43.7,"rank":35},{"arena":"agents","category":"mobileapps","elo":1164,"win_rate":52.9,"rank":28},{"arena":"agents","category":"webapps","elo":1073,"win_rate":43.2,"rank":34}],"artificial_analysis":{"intelligence_index":21.2,"coding_index":52.1,"agentic_index":17.5}},"reasoning":{"mandatory":false}},{"id":"anthropic/claude-sonnet-4.5:batch","canonical_slug":"anthropic/claude-4.5-sonnet-20250929","hugging_face_id":"","name":"Anthropic: Claude Sonnet 4.5 (batch)","created":1759161676,"description":"Claude Sonnet 4.5 is Anthropic’s most advanced Sonnet model to date, optimized for real-world agents and coding workflows. It delivers state-of-the-art performance on coding benchmarks such as SWE-bench Verified, with...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.0000015","completion":"0.0000075","web_search":"0.01","input_cache_read":"0.00000015","input_cache_write":"0.000001875","input_cache_write_1h":"0.000003","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000003","completion":"0.00001125","input_cache_read":"0.0000003","input_cache_write":"0.00000375","input_cache_write_1h":"0.000006"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":64000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":1,"top_p":1,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.5-sonnet-20250929/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1186,"win_rate":51,"rank":54},{"arena":"models","category":"asciiart","elo":1223,"win_rate":56.1,"rank":20},{"arena":"models","category":"codecategories","elo":1198,"win_rate":51.5,"rank":53},{"arena":"models","category":"dataviz","elo":1183,"win_rate":47.4,"rank":62},{"arena":"models","category":"gamedev","elo":1189,"win_rate":51.1,"rank":56},{"arena":"models","category":"svg","elo":1140,"win_rate":52.2,"rank":51},{"arena":"models","category":"uicomponent","elo":1188,"win_rate":49.4,"rank":57},{"arena":"models","category":"website","elo":1202,"win_rate":51.8,"rank":57},{"arena":"agents","category":"fullstack","elo":1059,"win_rate":43.7,"rank":35},{"arena":"agents","category":"mobileapps","elo":1164,"win_rate":52.9,"rank":28},{"arena":"agents","category":"webapps","elo":1073,"win_rate":43.2,"rank":34}],"artificial_analysis":{"intelligence_index":21.2,"coding_index":52.1,"agentic_index":17.5}},"reasoning":{"mandatory":false}},{"id":"deepseek/deepseek-v3.2-exp","canonical_slug":"deepseek/deepseek-v3.2-exp","hugging_face_id":"deepseek-ai/DeepSeek-V3.2-Exp","name":"DeepSeek: DeepSeek V3.2 Exp","created":1759150481,"description":"DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...","context_length":163840,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":"deepseek-v3.1"},"pricing":{"prompt":"0.00000027","completion":"0.00000041"},"top_provider":{"context_length":163840,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.6,"top_p":0.95,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-07-31","expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v3.2-exp/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1177,"win_rate":56.4,"rank":56},{"arena":"models","category":"codecategories","elo":1186,"win_rate":54.2,"rank":61},{"arena":"models","category":"dataviz","elo":1168,"win_rate":50.6,"rank":70},{"arena":"models","category":"gamedev","elo":1167,"win_rate":53,"rank":63},{"arena":"models","category":"svg","elo":1056,"win_rate":42,"rank":71},{"arena":"models","category":"uicomponent","elo":1182,"win_rate":53.3,"rank":59},{"arena":"models","category":"website","elo":1191,"win_rate":54.3,"rank":63}]},"reasoning":{"mandatory":false}},{"id":"thedrummer/cydonia-24b-v4.1","canonical_slug":"thedrummer/cydonia-24b-v4.1","hugging_face_id":"thedrummer/cydonia-24b-v4.1","name":"TheDrummer: Cydonia 24B V4.1","created":1758931878,"description":"Uncensored and creative writing model based on Mistral Small 3.2 24B with good recall, prompt adherence, and intelligence.","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000005","input_cache_read":"0.00000015"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-04-30","expiration_date":null,"links":{"details":"/api/v1/models/thedrummer/cydonia-24b-v4.1/endpoints"}},{"id":"relace/relace-apply-3","canonical_slug":"relace/relace-apply-3","hugging_face_id":"","name":"Relace: Relace Apply 3","created":1758891572,"description":"Relace Apply 3 is a specialized code-patching LLM that merges AI-suggested edits straight into your source files. It can apply updates from GPT-4o, Claude, and others into your files at...","context_length":256000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000085","completion":"0.00000125"},"top_provider":{"context_length":256000,"max_completion_tokens":128000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_tokens","seed","stop"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/relace/relace-apply-3/endpoints"}},{"id":"qwen/qwen3-vl-235b-a22b-thinking","canonical_slug":"qwen/qwen3-vl-235b-a22b-thinking","hugging_face_id":"Qwen/Qwen3-VL-235B-A22B-Thinking","name":"Qwen: Qwen3 VL 235B A22B Thinking","created":1758668690,"description":"Qwen3-VL-235B-A22B Thinking is a multimodal model that unifies strong text generation with visual understanding across images and video. The Thinking model is optimized for multimodal reasoning in STEM and math....","context_length":131072,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.0000004","completion":"0.000004"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.8,"top_p":0.95,"top_k":20,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":1},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-vl-235b-a22b-thinking/endpoints"},"reasoning":{"mandatory":true}},{"id":"qwen/qwen3-vl-235b-a22b-instruct","canonical_slug":"qwen/qwen3-vl-235b-a22b-instruct","hugging_face_id":"Qwen/Qwen3-VL-235B-A22B-Instruct","name":"Qwen: Qwen3 VL 235B A22B Instruct","created":1758668687,"description":"Qwen3-VL-235B-A22B Instruct is an open-weight multimodal model that unifies strong text generation with visual understanding across images and video. The Instruct model targets general vision-language use (VQA, document parsing, chart/table...","context_length":262144,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000021","completion":"0.0000019","input_cache_read":"0.0000001"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.7,"top_p":0.8,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-vl-235b-a22b-instruct/endpoints"}},{"id":"qwen/qwen3-max","canonical_slug":"qwen/qwen3-max","hugging_face_id":"","name":"Qwen: Qwen3 Max","created":1758662808,"description":"Qwen3-Max is an updated release built on the Qwen3 series, offering major improvements in reasoning, instruction following, multilingual support, and long-tail knowledge coverage compared to the January 2025 version. It...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000078","completion":"0.0000039","input_cache_read":"0.000000156","input_cache_write":"0.000000975","overrides":[{"min_prompt_tokens":32000,"prompt":"0.00000156","completion":"0.0000078","input_cache_read":"0.000000312","input_cache_write":"0.00000195"},{"min_prompt_tokens":128000,"prompt":"0.00000195","completion":"0.00000975","input_cache_read":"0.00000039","input_cache_write":"0.0000024375"}]},"top_provider":{"context_length":262144,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":1,"top_p":1,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-max/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1102,"win_rate":43.5,"rank":87},{"arena":"models","category":"asciiart","elo":1146,"win_rate":47.2,"rank":48},{"arena":"models","category":"codecategories","elo":1125,"win_rate":44,"rank":88},{"arena":"models","category":"dataviz","elo":1116,"win_rate":41.2,"rank":87},{"arena":"models","category":"gamedev","elo":1117,"win_rate":43.8,"rank":84},{"arena":"models","category":"svg","elo":1036,"win_rate":37.3,"rank":77},{"arena":"models","category":"uicomponent","elo":1093,"win_rate":40.2,"rank":89},{"arena":"models","category":"website","elo":1131,"win_rate":44.5,"rank":89}]},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3-coder-plus","canonical_slug":"qwen/qwen3-coder-plus","hugging_face_id":"","name":"Qwen: Qwen3 Coder Plus","created":1758662707,"description":"Qwen3 Coder Plus is Alibaba's proprietary version of the Open Source Qwen3 Coder 480B A35B. It is a powerful coding agent model specializing in autonomous programming via tool calling and...","context_length":1000000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000065","completion":"0.00000325","input_cache_read":"0.00000013","input_cache_write":"0.0000008125","overrides":[{"min_prompt_tokens":32000,"prompt":"0.00000117","completion":"0.00000585","input_cache_read":"0.000000234","input_cache_write":"0.0000014625"},{"min_prompt_tokens":128000,"prompt":"0.00000195","completion":"0.00000975","input_cache_read":"0.00000039","input_cache_write":"0.0000024375"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-coder-plus/endpoints"}},{"id":"deepseek/deepseek-v3.1-terminus","canonical_slug":"deepseek/deepseek-v3.1-terminus","hugging_face_id":"deepseek-ai/DeepSeek-V3.1-Terminus","name":"DeepSeek: DeepSeek V3.1 Terminus","created":1758548275,"description":"DeepSeek-V3.1 Terminus is an update to [DeepSeek V3.1](/deepseek/deepseek-chat-v3.1) that maintains the model's original capabilities while addressing issues reported by users, including language consistency and agent capabilities, further optimizing the model's...","context_length":163840,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":"deepseek-v3.1"},"pricing":{"prompt":"0.00000027","completion":"0.000001","input_cache_read":"0.000000135"},"top_provider":{"context_length":131072,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-v3.1-terminus/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1169,"win_rate":56,"rank":58},{"arena":"models","category":"codecategories","elo":1190,"win_rate":56,"rank":59},{"arena":"models","category":"dataviz","elo":1184,"win_rate":53.4,"rank":61},{"arena":"models","category":"gamedev","elo":1153,"win_rate":52.5,"rank":72},{"arena":"models","category":"svg","elo":1090,"win_rate":50.1,"rank":63},{"arena":"models","category":"uicomponent","elo":1198,"win_rate":59.5,"rank":52},{"arena":"models","category":"website","elo":1199,"win_rate":56.4,"rank":59}],"artificial_analysis":{"intelligence_index":15.4,"coding_index":43.5,"agentic_index":8.9}},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3-coder-flash","canonical_slug":"qwen/qwen3-coder-flash","hugging_face_id":"","name":"Qwen: Qwen3 Coder Flash","created":1758115536,"description":"Qwen3 Coder Flash is Alibaba's fast and cost efficient version of their proprietary Qwen3 Coder Plus. It is a powerful coding agent model specializing in autonomous programming via tool calling...","context_length":1000000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.000000195","completion":"0.000000975","input_cache_read":"0.000000039","input_cache_write":"0.00000024375","overrides":[{"min_prompt_tokens":32000,"prompt":"0.000000325","completion":"0.000001625","input_cache_read":"0.000000065","input_cache_write":"0.00000040625"},{"min_prompt_tokens":128000,"prompt":"0.00000052","completion":"0.0000026","input_cache_read":"0.000000104","input_cache_write":"0.00000065"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logprobs","max_tokens","presence_penalty","response_format","seed","stop","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-coder-flash/endpoints"}},{"id":"qwen/qwen3-next-80b-a3b-thinking","canonical_slug":"qwen/qwen3-next-80b-a3b-thinking-2509","hugging_face_id":"Qwen/Qwen3-Next-80B-A3B-Thinking","name":"Qwen: Qwen3 Next 80B A3B Thinking","created":1757612284,"description":"Qwen3-Next-80B-A3B-Thinking is a reasoning-first chat model in the Qwen3-Next line that outputs structured “thinking” traces by default. It’s designed for hard multi-step problems; math proofs, code synthesis/debugging, logic, and agentic...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000012"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-09-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-next-80b-a3b-thinking-2509/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":17.4,"agentic_index":null}},"reasoning":{"mandatory":true}},{"id":"qwen/qwen3-next-80b-a3b-instruct","canonical_slug":"qwen/qwen3-next-80b-a3b-instruct-2509","hugging_face_id":"Qwen/Qwen3-Next-80B-A3B-Instruct","name":"Qwen: Qwen3 Next 80B A3B Instruct","created":1757612213,"description":"Qwen3-Next-80B-A3B-Instruct is an instruction-tuned chat model in the Qwen3-Next series optimized for fast, stable responses without “thinking” traces. It targets complex tasks across reasoning, code generation, knowledge QA, and multilingual...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000009","completion":"0.0000011"},"top_provider":{"context_length":262144,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-09-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-next-80b-a3b-instruct-2509/endpoints"}},{"id":"qwen/qwen-plus-2025-07-28","canonical_slug":"qwen/qwen-plus-2025-07-28","hugging_face_id":"","name":"Qwen: Qwen Plus 0728","created":1757347599,"description":"Qwen Plus 0728, based on the Qwen3 foundation model, is a 1 million context hybrid reasoning model with a balanced performance, speed, and cost combination.","context_length":1000000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000026","completion":"0.00000078","overrides":[{"min_prompt_tokens":256000,"prompt":"0.00000078","completion":"0.00000234"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen-plus-2025-07-28/endpoints"},"reasoning":{"mandatory":false}},{"id":"moonshotai/kimi-k2-0905","canonical_slug":"moonshotai/kimi-k2-0905","hugging_face_id":"moonshotai/Kimi-K2-Instruct-0905","name":"MoonshotAI: Kimi K2 0905","created":1757021147,"description":"Kimi K2 0905 is the September update of [Kimi K2 0711](moonshotai/kimi-k2). It is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000006","completion":"0.0000025"},"top_provider":{"context_length":262144,"max_completion_tokens":98304,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-12-31","expiration_date":null,"links":{"details":"/api/v1/models/moonshotai/kimi-k2-0905/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"codecategories","elo":1113,"win_rate":48.5,"rank":91},{"arena":"models","category":"website","elo":1119,"win_rate":48.3,"rank":93}]}},{"id":"qwen/qwen3-30b-a3b-thinking-2507","canonical_slug":"qwen/qwen3-30b-a3b-thinking-2507","hugging_face_id":"Qwen/Qwen3-30B-A3B-Thinking-2507","name":"Qwen: Qwen3 30B A3B Thinking 2507","created":1756399192,"description":"Qwen3-30B-A3B-Thinking-2507 is a 30B parameter Mixture-of-Experts reasoning model optimized for complex tasks requiring extended multi-step thinking. The model is designed specifically for “thinking mode,” where internal reasoning traces are separated...","context_length":81920,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.0000024"},"top_provider":{"context_length":81920,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","response_format","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-30b-a3b-thinking-2507/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"dataviz","elo":940,"win_rate":33.3,"rank":114},{"arena":"models","category":"website","elo":942,"win_rate":35.5,"rank":122}],"artificial_analysis":{"intelligence_index":9.8,"coding_index":12.1,"agentic_index":0.9}},"reasoning":{"mandatory":true}},{"id":"nousresearch/hermes-4-405b","canonical_slug":"nousresearch/hermes-4-405b","hugging_face_id":"NousResearch/Hermes-4-405B","name":"Nous: Hermes 4 405B","created":1756235463,"description":"Hermes 4 is a large-scale reasoning model built on Meta-Llama-3.1-405B and released by Nous Research. It introduces a hybrid reasoning mode, where the model can choose to deliberate internally with...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000003"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","repetition_penalty","response_format","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-08-31","expiration_date":null,"links":{"details":"/api/v1/models/nousresearch/hermes-4-405b/endpoints"},"reasoning":{"mandatory":false}},{"id":"deepseek/deepseek-chat-v3.1","canonical_slug":"deepseek/deepseek-chat-v3.1","hugging_face_id":"deepseek-ai/DeepSeek-V3.1","name":"DeepSeek: DeepSeek V3.1","created":1755779628,"description":"DeepSeek-V3.1 is a large hybrid reasoning model (671B parameters, 37B active) that supports both thinking and non-thinking modes via prompt templates. It extends the DeepSeek-V3 base with a two-phase long-context...","context_length":163840,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":"deepseek-v3.1"},"pricing":{"prompt":"0.00000025","completion":"0.00000095","input_cache_read":"0.00000013"},"top_provider":{"context_length":163840,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-chat-v3.1/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1105,"win_rate":47.9,"rank":86},{"arena":"models","category":"codecategories","elo":1127,"win_rate":47.8,"rank":87},{"arena":"models","category":"dataviz","elo":1112,"win_rate":46.2,"rank":91},{"arena":"models","category":"gamedev","elo":1107,"win_rate":47.1,"rank":86},{"arena":"models","category":"svg","elo":992,"win_rate":38.2,"rank":84},{"arena":"models","category":"uicomponent","elo":1100,"win_rate":47.1,"rank":87},{"arena":"models","category":"website","elo":1135,"win_rate":48,"rank":86}]},"reasoning":{"mandatory":false}},{"id":"mistralai/mistral-medium-3.1","canonical_slug":"mistralai/mistral-medium-3.1","hugging_face_id":"","name":"Mistral: Mistral Medium 3.1","created":1755095639,"description":"Mistral Medium 3.1 is an updated version of Mistral Medium 3, which is a high-performance enterprise-grade language model designed to deliver frontier-level capabilities at significantly reduced operational cost. It balances...","context_length":131072,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.0000004","completion":"0.000002","input_cache_read":"0.00000004"},"top_provider":{"context_length":131072,"max_completion_tokens":104857,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2025-06-30","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-medium-3.1/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1110,"win_rate":44.6,"rank":82},{"arena":"models","category":"asciiart","elo":1016,"win_rate":30.8,"rank":65},{"arena":"models","category":"codecategories","elo":1135,"win_rate":45.1,"rank":83},{"arena":"models","category":"dataviz","elo":1160,"win_rate":47.2,"rank":72},{"arena":"models","category":"gamedev","elo":1097,"win_rate":40.7,"rank":90},{"arena":"models","category":"svg","elo":1017,"win_rate":38.2,"rank":80},{"arena":"models","category":"uicomponent","elo":1115,"win_rate":43.6,"rank":83},{"arena":"models","category":"website","elo":1145,"win_rate":46,"rank":83}],"artificial_analysis":{"intelligence_index":null,"coding_index":20.5,"agentic_index":3.1}}},{"id":"mistralai/mistral-medium-3.1:batch","canonical_slug":"mistralai/mistral-medium-3.1","hugging_face_id":"","name":"Mistral: Mistral Medium 3.1 (batch)","created":1755095639,"description":"Mistral Medium 3.1 is an updated version of Mistral Medium 3, which is a high-performance enterprise-grade language model designed to deliver frontier-level capabilities at significantly reduced operational cost. It balances...","context_length":131072,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.000001","input_cache_read":"0.00000002"},"top_provider":{"context_length":131072,"max_completion_tokens":104857,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2025-06-30","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-medium-3.1/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1110,"win_rate":44.6,"rank":82},{"arena":"models","category":"asciiart","elo":1016,"win_rate":30.8,"rank":65},{"arena":"models","category":"codecategories","elo":1135,"win_rate":45.1,"rank":83},{"arena":"models","category":"dataviz","elo":1160,"win_rate":47.2,"rank":72},{"arena":"models","category":"gamedev","elo":1097,"win_rate":40.7,"rank":90},{"arena":"models","category":"svg","elo":1017,"win_rate":38.2,"rank":80},{"arena":"models","category":"uicomponent","elo":1115,"win_rate":43.6,"rank":83},{"arena":"models","category":"website","elo":1145,"win_rate":46,"rank":83}],"artificial_analysis":{"intelligence_index":null,"coding_index":20.5,"agentic_index":3.1}}},{"id":"z-ai/glm-4.5v","canonical_slug":"z-ai/glm-4.5v","hugging_face_id":"zai-org/GLM-4.5V","name":"Z.ai: GLM 4.5V","created":1754922288,"description":"GLM-4.5V is a vision-language foundation model for multimodal agent applications. Built on a Mixture-of-Experts (MoE) architecture with 106B parameters and 12B activated parameters, it achieves state-of-the-art results in video understanding,...","context_length":65536,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000006","completion":"0.0000018","input_cache_read":"0.00000011"},"top_provider":{"context_length":65536,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":0.75,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-12-31","expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-4.5v/endpoints"},"reasoning":{"mandatory":false}},{"id":"openai/gpt-5","canonical_slug":"openai/gpt-5-2025-08-07","hugging_face_id":"","name":"OpenAI: GPT-5","created":1754587413,"description":"GPT-5 is OpenAI’s most advanced model, offering major improvements in reasoning, code quality, and user experience. It is optimized for complex tasks that require step-by-step reasoning, instruction following, and accuracy...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.00001","web_search":"0.01","input_cache_read":"0.000000125"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-09-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5-2025-08-07/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1084,"win_rate":41.3,"rank":91},{"arena":"models","category":"asciiart","elo":1159,"win_rate":49,"rank":40},{"arena":"models","category":"codecategories","elo":1183,"win_rate":54.5,"rank":64},{"arena":"models","category":"dataviz","elo":1237,"win_rate":60.5,"rank":40},{"arena":"models","category":"gamedev","elo":1208,"win_rate":59.3,"rank":47},{"arena":"models","category":"svg","elo":1213,"win_rate":64.1,"rank":25},{"arena":"models","category":"uicomponent","elo":1195,"win_rate":57.4,"rank":55},{"arena":"models","category":"website","elo":1197,"win_rate":53.7,"rank":61}],"artificial_analysis":{"intelligence_index":null,"coding_index":37.8,"agentic_index":null}},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"openai/gpt-5:batch","canonical_slug":"openai/gpt-5-2025-08-07","hugging_face_id":"","name":"OpenAI: GPT-5 (batch)","created":1754587413,"description":"GPT-5 is OpenAI’s most advanced model, offering major improvements in reasoning, code quality, and user experience. It is optimized for complex tasks that require step-by-step reasoning, instruction following, and accuracy...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000000625","completion":"0.000005","web_search":"0.01","input_cache_read":"0.0000000625"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-09-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5-2025-08-07/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1084,"win_rate":41.3,"rank":91},{"arena":"models","category":"asciiart","elo":1159,"win_rate":49,"rank":40},{"arena":"models","category":"codecategories","elo":1183,"win_rate":54.5,"rank":64},{"arena":"models","category":"dataviz","elo":1237,"win_rate":60.5,"rank":40},{"arena":"models","category":"gamedev","elo":1208,"win_rate":59.3,"rank":47},{"arena":"models","category":"svg","elo":1213,"win_rate":64.1,"rank":25},{"arena":"models","category":"uicomponent","elo":1195,"win_rate":57.4,"rank":55},{"arena":"models","category":"website","elo":1197,"win_rate":53.7,"rank":61}],"artificial_analysis":{"intelligence_index":null,"coding_index":37.8,"agentic_index":null}},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"openai/gpt-5-mini","canonical_slug":"openai/gpt-5-mini-2025-08-07","hugging_face_id":"","name":"OpenAI: GPT-5 Mini","created":1754587407,"description":"GPT-5 Mini is a compact version of GPT-5, designed to handle lighter-weight reasoning tasks. It provides the same instruction-following and safety-tuning benefits as GPT-5, but with reduced latency and cost....","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.000002","web_search":"0.01","input_cache_read":"0.000000025"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-05-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5-mini-2025-08-07/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1065,"win_rate":36.9,"rank":94},{"arena":"models","category":"asciiart","elo":1140,"win_rate":44.5,"rank":49},{"arena":"models","category":"codecategories","elo":1130,"win_rate":43.5,"rank":86},{"arena":"models","category":"dataviz","elo":1146,"win_rate":43.7,"rank":78},{"arena":"models","category":"gamedev","elo":1152,"win_rate":46.5,"rank":73},{"arena":"models","category":"svg","elo":1116,"win_rate":45.8,"rank":57},{"arena":"models","category":"uicomponent","elo":1121,"win_rate":41.9,"rank":79},{"arena":"models","category":"website","elo":1137,"win_rate":44.3,"rank":85}],"artificial_analysis":{"intelligence_index":17.4,"coding_index":15.6,"agentic_index":8.9}},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"openai/gpt-5-mini:batch","canonical_slug":"openai/gpt-5-mini-2025-08-07","hugging_face_id":"","name":"OpenAI: GPT-5 Mini (batch)","created":1754587407,"description":"GPT-5 Mini is a compact version of GPT-5, designed to handle lighter-weight reasoning tasks. It provides the same instruction-following and safety-tuning benefits as GPT-5, but with reduced latency and cost....","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000000125","completion":"0.000001","web_search":"0.01","input_cache_read":"0.0000000125"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-05-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5-mini-2025-08-07/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1065,"win_rate":36.9,"rank":94},{"arena":"models","category":"asciiart","elo":1140,"win_rate":44.5,"rank":49},{"arena":"models","category":"codecategories","elo":1130,"win_rate":43.5,"rank":86},{"arena":"models","category":"dataviz","elo":1146,"win_rate":43.7,"rank":78},{"arena":"models","category":"gamedev","elo":1152,"win_rate":46.5,"rank":73},{"arena":"models","category":"svg","elo":1116,"win_rate":45.8,"rank":57},{"arena":"models","category":"uicomponent","elo":1121,"win_rate":41.9,"rank":79},{"arena":"models","category":"website","elo":1137,"win_rate":44.3,"rank":85}],"artificial_analysis":{"intelligence_index":17.4,"coding_index":15.6,"agentic_index":8.9}},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"openai/gpt-5-nano","canonical_slug":"openai/gpt-5-nano-2025-08-07","hugging_face_id":"","name":"OpenAI: GPT-5 Nano","created":1754587402,"description":"GPT-5-Nano is the smallest and fastest variant in the GPT-5 system, optimized for developer tools, rapid interactions, and ultra-low latency environments. While limited in reasoning depth compared to its larger...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000005","completion":"0.0000004","web_search":"0.01","input_cache_read":"0.000000005"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_completion_tokens","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-05-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5-nano-2025-08-07/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":994,"win_rate":36.3,"rank":106},{"arena":"models","category":"codecategories","elo":1099,"win_rate":48,"rank":94},{"arena":"models","category":"dataviz","elo":1071,"win_rate":46.2,"rank":98},{"arena":"models","category":"gamedev","elo":1068,"win_rate":46.5,"rank":96},{"arena":"models","category":"uicomponent","elo":1082,"win_rate":51.9,"rank":92},{"arena":"models","category":"website","elo":1113,"win_rate":48.9,"rank":94}]},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"openai/gpt-5-nano:batch","canonical_slug":"openai/gpt-5-nano-2025-08-07","hugging_face_id":"","name":"OpenAI: GPT-5 Nano (batch)","created":1754587402,"description":"GPT-5-Nano is the smallest and fastest variant in the GPT-5 system, optimized for developer tools, rapid interactions, and ultra-low latency environments. While limited in reasoning depth compared to its larger...","context_length":400000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000000025","completion":"0.0000002","web_search":"0.01","input_cache_read":"0.0000000025"},"top_provider":{"context_length":400000,"max_completion_tokens":128000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-05-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-5-nano-2025-08-07/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":994,"win_rate":36.3,"rank":106},{"arena":"models","category":"codecategories","elo":1099,"win_rate":48,"rank":94},{"arena":"models","category":"dataviz","elo":1071,"win_rate":46.2,"rank":98},{"arena":"models","category":"gamedev","elo":1068,"win_rate":46.5,"rank":96},{"arena":"models","category":"uicomponent","elo":1082,"win_rate":51.9,"rank":92},{"arena":"models","category":"website","elo":1113,"win_rate":48.9,"rank":94}]},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low","minimal"],"default_effort":"medium"}},{"id":"openai/gpt-oss-120b","canonical_slug":"openai/gpt-oss-120b","hugging_face_id":"openai/gpt-oss-120b","name":"OpenAI: gpt-oss-120b","created":1754414231,"description":"gpt-oss-120b is an open-weight, 117B-parameter Mixture-of-Experts (MoE) language model from OpenAI designed for high-reasoning, agentic, and general-purpose production use cases. It activates 5.1B parameters per forward pass and is optimized...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000000037","completion":"0.00000017"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_a","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-oss-120b/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":929,"win_rate":29.4,"rank":109},{"arena":"models","category":"codecategories","elo":978,"win_rate":33.4,"rank":117},{"arena":"models","category":"dataviz","elo":1003,"win_rate":43.6,"rank":107},{"arena":"models","category":"gamedev","elo":1015,"win_rate":40.5,"rank":106},{"arena":"models","category":"uicomponent","elo":942,"win_rate":35.7,"rank":112},{"arena":"models","category":"website","elo":980,"win_rate":32.5,"rank":118}],"artificial_analysis":{"intelligence_index":12.3,"coding_index":30.4,"agentic_index":6.2}},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"openai/gpt-oss-120b:batch","canonical_slug":"openai/gpt-oss-120b","hugging_face_id":"openai/gpt-oss-120b","name":"OpenAI: gpt-oss-120b (batch)","created":1754414231,"description":"gpt-oss-120b is an open-weight, 117B-parameter Mixture-of-Experts (MoE) language model from OpenAI designed for high-reasoning, agentic, and general-purpose production use cases. It activates 5.1B parameters per forward pass and is optimized...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000006"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-oss-120b/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":929,"win_rate":29.4,"rank":109},{"arena":"models","category":"codecategories","elo":978,"win_rate":33.4,"rank":117},{"arena":"models","category":"dataviz","elo":1003,"win_rate":43.6,"rank":107},{"arena":"models","category":"gamedev","elo":1015,"win_rate":40.5,"rank":106},{"arena":"models","category":"uicomponent","elo":942,"win_rate":35.7,"rank":112},{"arena":"models","category":"website","elo":980,"win_rate":32.5,"rank":118}],"artificial_analysis":{"intelligence_index":12.3,"coding_index":30.4,"agentic_index":6.2}},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"openai/gpt-oss-20b","canonical_slug":"openai/gpt-oss-20b","hugging_face_id":"openai/gpt-oss-20b","name":"OpenAI: gpt-oss-20b","created":1754414229,"description":"gpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000003","completion":"0.00000013","input_cache_read":"0.00000003"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-oss-20b/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"dataviz","elo":947,"win_rate":39.7,"rank":112},{"arena":"models","category":"website","elo":865,"win_rate":27.9,"rank":126}],"artificial_analysis":{"intelligence_index":9,"coding_index":20.7,"agentic_index":1.4}},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"openai/gpt-oss-20b:batch","canonical_slug":"openai/gpt-oss-20b","hugging_face_id":"openai/gpt-oss-20b","name":"OpenAI: gpt-oss-20b (batch)","created":1754414229,"description":"gpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000005","completion":"0.0000002"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","stop","structured_outputs","temperature","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-oss-20b/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"dataviz","elo":947,"win_rate":39.7,"rank":112},{"arena":"models","category":"website","elo":865,"win_rate":27.9,"rank":126}],"artificial_analysis":{"intelligence_index":9,"coding_index":20.7,"agentic_index":1.4}},"reasoning":{"mandatory":true,"supported_efforts":["high","medium","low"],"default_effort":"medium"}},{"id":"anthropic/claude-opus-4.1","canonical_slug":"anthropic/claude-4.1-opus-20250805","hugging_face_id":"","name":"Anthropic: Claude Opus 4.1","created":1754411591,"description":"Claude Opus 4.1 is an updated version of Anthropic’s flagship model, offering improved performance in coding, reasoning, and agentic tasks. It achieves 74.5% on SWE-bench Verified and shows notable gains...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000015","completion":"0.000075","web_search":"0.01","input_cache_read":"0.0000015","input_cache_write":"0.00001875","input_cache_write_1h":"0.00003"},"top_provider":{"context_length":200000,"max_completion_tokens":32000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.1-opus-20250805/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1181,"win_rate":51.9,"rank":55},{"arena":"models","category":"asciiart","elo":1185,"win_rate":51.5,"rank":26},{"arena":"models","category":"codecategories","elo":1186,"win_rate":55.8,"rank":60},{"arena":"models","category":"dataviz","elo":1182,"win_rate":56.4,"rank":63},{"arena":"models","category":"gamedev","elo":1194,"win_rate":58.4,"rank":53},{"arena":"models","category":"svg","elo":1177,"win_rate":60.8,"rank":37},{"arena":"models","category":"uicomponent","elo":1180,"win_rate":57.8,"rank":61},{"arena":"models","category":"website","elo":1189,"win_rate":55.3,"rank":64}]},"reasoning":{"mandatory":false}},{"id":"anthropic/claude-opus-4.1:batch","canonical_slug":"anthropic/claude-4.1-opus-20250805","hugging_face_id":"","name":"Anthropic: Claude Opus 4.1 (batch)","created":1754411591,"description":"Claude Opus 4.1 is an updated version of Anthropic’s flagship model, offering improved performance in coding, reasoning, and agentic tasks. It achieves 74.5% on SWE-bench Verified and shows notable gains...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.0000075","completion":"0.0000375","web_search":"0.01","input_cache_read":"0.00000075","input_cache_write":"0.000009375","input_cache_write_1h":"0.000015"},"top_provider":{"context_length":200000,"max_completion_tokens":32000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","stop","structured_outputs","temperature","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4.1-opus-20250805/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1181,"win_rate":51.9,"rank":55},{"arena":"models","category":"asciiart","elo":1185,"win_rate":51.5,"rank":26},{"arena":"models","category":"codecategories","elo":1186,"win_rate":55.8,"rank":60},{"arena":"models","category":"dataviz","elo":1182,"win_rate":56.4,"rank":63},{"arena":"models","category":"gamedev","elo":1194,"win_rate":58.4,"rank":53},{"arena":"models","category":"svg","elo":1177,"win_rate":60.8,"rank":37},{"arena":"models","category":"uicomponent","elo":1180,"win_rate":57.8,"rank":61},{"arena":"models","category":"website","elo":1189,"win_rate":55.3,"rank":64}]},"reasoning":{"mandatory":false}},{"id":"mistralai/codestral-2508","canonical_slug":"mistralai/codestral-2508","hugging_face_id":"","name":"Mistral: Codestral 2508","created":1754079630,"description":"Mistral's cutting-edge language model for coding released end of July 2025. Codestral specializes in low-latency, high-frequency tasks such as fill-in-the-middle (FIM), code correction and test generation.\n\n[Blog Post](https://mistral.ai/news/codestral-25-08)","context_length":256000,"architecture":{"modality":"text+file->text","input_modalities":["text","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000009","input_cache_read":"0.00000003"},"top_provider":{"context_length":256000,"max_completion_tokens":204800,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","prediction","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/codestral-2508/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"codecategories","elo":1022,"win_rate":38.5,"rank":111},{"arena":"models","category":"dataviz","elo":1033,"win_rate":41.1,"rank":102},{"arena":"models","category":"gamedev","elo":990,"win_rate":36.3,"rank":112},{"arena":"models","category":"uicomponent","elo":1032,"win_rate":46.7,"rank":101},{"arena":"models","category":"website","elo":1025,"win_rate":37.8,"rank":113},{"arena":"models","category":"3d","elo":1048,"win_rate":45.4,"rank":97}]}},{"id":"mistralai/codestral-2508:batch","canonical_slug":"mistralai/codestral-2508","hugging_face_id":"","name":"Mistral: Codestral 2508 (batch)","created":1754079630,"description":"Mistral's cutting-edge language model for coding released end of July 2025. Codestral specializes in low-latency, high-frequency tasks such as fill-in-the-middle (FIM), code correction and test generation.\n\n[Blog Post](https://mistral.ai/news/codestral-25-08)","context_length":256000,"architecture":{"modality":"text+file->text","input_modalities":["text","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.00000045","input_cache_read":"0.000000015"},"top_provider":{"context_length":256000,"max_completion_tokens":204800,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","prediction","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/codestral-2508/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"codecategories","elo":1022,"win_rate":38.5,"rank":111},{"arena":"models","category":"dataviz","elo":1033,"win_rate":41.1,"rank":102},{"arena":"models","category":"gamedev","elo":990,"win_rate":36.3,"rank":112},{"arena":"models","category":"uicomponent","elo":1032,"win_rate":46.7,"rank":101},{"arena":"models","category":"website","elo":1025,"win_rate":37.8,"rank":113},{"arena":"models","category":"3d","elo":1048,"win_rate":45.4,"rank":97}]}},{"id":"qwen/qwen3-coder-30b-a3b-instruct","canonical_slug":"qwen/qwen3-coder-30b-a3b-instruct","hugging_face_id":"Qwen/Qwen3-Coder-30B-A3B-Instruct","name":"Qwen: Qwen3 Coder 30B A3B Instruct","created":1753972379,"description":"Qwen3-Coder-30B-A3B-Instruct is a 30.5B parameter Mixture-of-Experts (MoE) model with 128 experts (8 active per forward pass), designed for advanced code generation, repository-scale understanding, and agentic tool use. Built on the...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000007","completion":"0.00000028"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logprobs","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-coder-30b-a3b-instruct/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"dataviz","elo":1099,"win_rate":54.7,"rank":93},{"arena":"models","category":"uicomponent","elo":1061,"win_rate":54.1,"rank":93},{"arena":"models","category":"website","elo":1099,"win_rate":57.1,"rank":97}]}},{"id":"qwen/qwen3-30b-a3b-instruct-2507","canonical_slug":"qwen/qwen3-30b-a3b-instruct-2507","hugging_face_id":"Qwen/Qwen3-30B-A3B-Instruct-2507","name":"Qwen: Qwen3 30B A3B Instruct 2507","created":1753806965,"description":"Qwen3-30B-A3B-Instruct-2507 is a 30.5B-parameter mixture-of-experts language model from Qwen, with 3.3B active parameters per inference. It operates in non-thinking mode and is designed for high-quality instruction following, multilingual understanding, and...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.00000004815","completion":"0.00000019305"},"top_provider":{"context_length":128000,"max_completion_tokens":32000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-30b-a3b-instruct-2507/endpoints"}},{"id":"z-ai/glm-4.5","canonical_slug":"z-ai/glm-4.5","hugging_face_id":"zai-org/GLM-4.5","name":"Z.ai: GLM 4.5","created":1753471347,"description":"GLM-4.5 is our latest flagship foundation model, purpose-built for agent-based applications. It leverages a Mixture-of-Experts (MoE) architecture and supports a context length of up to 128k tokens. GLM-4.5 delivers significantly...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000006","completion":"0.0000022","input_cache_read":"0.00000011"},"top_provider":{"context_length":131072,"max_completion_tokens":98304,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":0.75,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-12-31","expiration_date":"2026-12-31","links":{"details":"/api/v1/models/z-ai/glm-4.5/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1199,"win_rate":59.7,"rank":48},{"arena":"models","category":"codecategories","elo":1180,"win_rate":54.4,"rank":66},{"arena":"models","category":"dataviz","elo":1175,"win_rate":53.3,"rank":68},{"arena":"models","category":"gamedev","elo":1168,"win_rate":54.4,"rank":62},{"arena":"models","category":"svg","elo":1121,"win_rate":50.8,"rank":56},{"arena":"models","category":"uicomponent","elo":1159,"win_rate":55,"rank":69},{"arena":"models","category":"website","elo":1182,"win_rate":53.7,"rank":68}]},"reasoning":{"mandatory":false}},{"id":"z-ai/glm-4.5-air","canonical_slug":"z-ai/glm-4.5-air","hugging_face_id":"zai-org/GLM-4.5-Air","name":"Z.ai: GLM 4.5 Air","created":1753471258,"description":"GLM-4.5-Air is the lightweight variant of our latest flagship model family, also purpose-built for agent-centric applications. Like GLM-4.5, it adopts the Mixture-of-Experts (MoE) architecture but with a more compact parameter...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000013","completion":"0.00000085","input_cache_read":"0.000000025"},"top_provider":{"context_length":131072,"max_completion_tokens":98304,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","repetition_penalty","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":0.75,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-12-31","expiration_date":null,"links":{"details":"/api/v1/models/z-ai/glm-4.5-air/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1151,"win_rate":54.1,"rank":67},{"arena":"models","category":"codecategories","elo":1152,"win_rate":51.4,"rank":79},{"arena":"models","category":"dataviz","elo":1202,"win_rate":58.5,"rank":51},{"arena":"models","category":"gamedev","elo":1117,"win_rate":48.4,"rank":83},{"arena":"models","category":"svg","elo":1096,"win_rate":50.8,"rank":62},{"arena":"models","category":"uicomponent","elo":1138,"win_rate":54.4,"rank":75},{"arena":"models","category":"website","elo":1159,"win_rate":51.3,"rank":76}]},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3-235b-a22b-thinking-2507","canonical_slug":"qwen/qwen3-235b-a22b-thinking-2507","hugging_face_id":"Qwen/Qwen3-235B-A22B-Thinking-2507","name":"Qwen: Qwen3 235B A22B Thinking 2507","created":1753449557,"description":"Qwen3-235B-A22B-Thinking-2507 is a high-performance, open-weight Mixture-of-Experts (MoE) language model optimized for complex reasoning tasks. It activates 22B of its 235B parameters per forward pass and natively supports up to 262,144...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":"qwen3"},"pricing":{"prompt":"0.00000023","completion":"0.0000023"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-235b-a22b-thinking-2507/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1026,"win_rate":40.4,"rank":100},{"arena":"models","category":"codecategories","elo":1048,"win_rate":40.8,"rank":105},{"arena":"models","category":"dataviz","elo":965,"win_rate":32.6,"rank":110},{"arena":"models","category":"gamedev","elo":980,"win_rate":34.2,"rank":113},{"arena":"models","category":"uicomponent","elo":960,"win_rate":34,"rank":110},{"arena":"models","category":"website","elo":1065,"win_rate":42,"rank":103}],"artificial_analysis":{"intelligence_index":12.7,"coding_index":22.1,"agentic_index":1.3}},"reasoning":{"mandatory":true}},{"id":"qwen/qwen3-coder","canonical_slug":"qwen/qwen3-coder-480b-a35b-07-25","hugging_face_id":"Qwen/Qwen3-Coder-480B-A35B-Instruct","name":"Qwen: Qwen3 Coder 480B A35B","created":1753230546,"description":"Qwen3-Coder-480B-A35B-Instruct is a Mixture-of-Experts (MoE) code generation model developed by the Qwen team. It is optimized for agentic coding tasks such as function calling, tool use, and long-context reasoning over...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.000001","input_cache_read":"0.0000001"},"top_provider":{"context_length":262144,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-coder-480b-a35b-07-25/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"codecategories","elo":1158,"win_rate":61.2,"rank":73},{"arena":"models","category":"dataviz","elo":1098,"win_rate":54.9,"rank":94},{"arena":"models","category":"gamedev","elo":1122,"win_rate":58.7,"rank":80},{"arena":"models","category":"uicomponent","elo":1128,"win_rate":61.4,"rank":77},{"arena":"models","category":"website","elo":1171,"win_rate":61.7,"rank":74}]}},{"id":"bytedance/ui-tars-1.5-7b","canonical_slug":"bytedance/ui-tars-1.5-7b","hugging_face_id":"ByteDance-Seed/UI-TARS-1.5-7B","name":"ByteDance: UI-TARS 7B ","created":1753205056,"description":"UI-TARS-1.5 is a multimodal vision-language agent optimized for GUI-based environments, including desktop interfaces, web browsers, mobile systems, and games. Built by ByteDance, it builds upon the UI-TARS framework with reinforcement...","context_length":128000,"architecture":{"modality":"text+image->text","input_modalities":["image","text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000002","input_cache_read":"0.0000001"},"top_provider":{"context_length":128000,"max_completion_tokens":2048,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","repetition_penalty","seed","stop","structured_outputs","temperature","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/bytedance/ui-tars-1.5-7b/endpoints"}},{"id":"google/gemini-2.5-flash-lite","canonical_slug":"google/gemini-2.5-flash-lite","hugging_face_id":"","name":"Google: Gemini 2.5 Flash Lite","created":1753200276,"description":"Gemini 2.5 Flash-Lite is a lightweight reasoning model in the Gemini 2.5 family, optimized for ultra-low latency and cost efficiency. It offers improved throughput, faster token generation, and better performance...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","file","audio","video"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000004","image":"0.0000001","audio":"0.0000003","input_audio_cache":"0.00000003","web_search":"0.014","internal_reasoning":"0.0000004","input_cache_read":"0.00000001","input_cache_write":"0.0000000833333333333333"},"top_provider":{"context_length":1048576,"max_completion_tokens":65535,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-2.5-flash-lite/endpoints"},"reasoning":{"mandatory":false}},{"id":"google/gemini-2.5-flash-lite:batch","canonical_slug":"google/gemini-2.5-flash-lite","hugging_face_id":"","name":"Google: Gemini 2.5 Flash Lite (batch)","created":1753200276,"description":"Gemini 2.5 Flash-Lite is a lightweight reasoning model in the Gemini 2.5 family, optimized for ultra-low latency and cost efficiency. It offers improved throughput, faster token generation, and better performance...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","file","audio","video"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000005","completion":"0.0000002","image":"0.00000005","audio":"0.00000015","input_audio_cache":"0.00000003","web_search":"0.014","internal_reasoning":"0.0000002","input_cache_read":"0.00000001"},"top_provider":{"context_length":1048576,"max_completion_tokens":65535,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-2.5-flash-lite/endpoints"},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3-235b-a22b-2507","canonical_slug":"qwen/qwen3-235b-a22b-07-25","hugging_face_id":"Qwen/Qwen3-235B-A22B-Instruct-2507","name":"Qwen: Qwen3 235B A22B Instruct 2507","created":1753119555,"description":"Qwen3-235B-A22B-Instruct-2507 is a multilingual, instruction-tuned mixture-of-experts language model based on the Qwen3-235B architecture, with 22B active parameters per forward pass. It is optimized for general-purpose text generation, including instruction following,...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":null},"pricing":{"prompt":"0.0000000875","completion":"0.00000035","input_cache_read":"0.0000000175"},"top_provider":{"context_length":262144,"max_completion_tokens":235929,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-235b-a22b-07-25/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1023,"win_rate":41.1,"rank":101},{"arena":"models","category":"codecategories","elo":1053,"win_rate":42.6,"rank":102},{"arena":"models","category":"dataviz","elo":1085,"win_rate":49,"rank":95},{"arena":"models","category":"gamedev","elo":974,"win_rate":35,"rank":114},{"arena":"models","category":"uicomponent","elo":980,"win_rate":38.4,"rank":106},{"arena":"models","category":"website","elo":1070,"win_rate":43.6,"rank":102}]}},{"id":"moonshotai/kimi-k2","canonical_slug":"moonshotai/kimi-k2","hugging_face_id":"moonshotai/Kimi-K2-Instruct","name":"MoonshotAI: Kimi K2 0711","created":1752263252,"description":"Kimi K2 Instruct is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32 billion active per forward pass. It is optimized for...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000057","completion":"0.0000023"},"top_provider":{"context_length":131072,"max_completion_tokens":98304,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","repetition_penalty","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-12-31","expiration_date":null,"links":{"details":"/api/v1/models/moonshotai/kimi-k2/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"codecategories","elo":1048,"win_rate":51.7,"rank":104},{"arena":"models","category":"dataviz","elo":1032,"win_rate":49.4,"rank":103},{"arena":"models","category":"gamedev","elo":997,"win_rate":46.4,"rank":110},{"arena":"models","category":"uicomponent","elo":1046,"win_rate":55.1,"rank":95},{"arena":"models","category":"website","elo":1062,"win_rate":53.1,"rank":105}]}},{"id":"cognitivecomputations/dolphin-mistral-24b-venice-edition","canonical_slug":"venice/uncensored","hugging_face_id":"cognitivecomputations/Dolphin-Mistral-24B-Venice-Edition","name":"Venice: Uncensored","created":1752094966,"description":"Venice Uncensored Dolphin Mistral 24B Venice Edition is a fine-tuned variant of Mistral-Small-24B-Instruct-2501, developed by dphn.ai in collaboration with Venice.ai. This model is designed as an “uncensored” instruct-tuned LLM, preserving...","context_length":128000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.0000009"},"top_provider":{"context_length":128000,"max_completion_tokens":8192,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","stop","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-04-30","expiration_date":null,"links":{"details":"/api/v1/models/venice/uncensored/endpoints"}},{"id":"tencent/hunyuan-a13b-instruct","canonical_slug":"tencent/hunyuan-a13b-instruct","hugging_face_id":"tencent/Hunyuan-A13B-Instruct","name":"Tencent: Hunyuan A13B Instruct","created":1751987664,"description":"Hunyuan-A13B is a 13B active parameter Mixture-of-Experts (MoE) language model developed by Tencent, with a total parameter count of 80B and support for reasoning via Chain-of-Thought. It offers competitive benchmark...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000014","completion":"0.00000057"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","reasoning","response_format","structured_outputs","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/tencent/hunyuan-a13b-instruct/endpoints"},"reasoning":{"mandatory":false}},{"id":"morph/morph-v3-large","canonical_slug":"morph/morph-v3-large","hugging_face_id":"","name":"Morph: Morph V3 Large","created":1751910858,"description":"Morph's high-accuracy apply model for complex code edits. ~4,500 tokens/sec with 98% accuracy for precise code transformations. The model requires the prompt to be in the following format: {instruction} {initial_code}...","context_length":262144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000009","completion":"0.0000019"},"top_provider":{"context_length":262144,"max_completion_tokens":131072,"is_moderated":false},"per_request_limits":null,"supported_parameters":["logprobs","max_tokens","response_format","stop","structured_outputs","temperature","top_logprobs"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/morph/morph-v3-large/endpoints"}},{"id":"morph/morph-v3-fast","canonical_slug":"morph/morph-v3-fast","hugging_face_id":"","name":"Morph: Morph V3 Fast","created":1751910002,"description":"Morph's fastest apply model for code edits. ~10,500 tokens/sec with 96% accuracy for rapid code transformations. The model requires the prompt to be in the following format: {instruction} {initial_code} {edit_snippet}...","context_length":81920,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000008","completion":"0.0000012"},"top_provider":{"context_length":81920,"max_completion_tokens":38000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_tokens","stop","temperature"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/morph/morph-v3-fast/endpoints"}},{"id":"baidu/ernie-4.5-vl-424b-a47b","canonical_slug":"baidu/ernie-4.5-vl-424b-a47b","hugging_face_id":"baidu/ERNIE-4.5-VL-424B-A47B-PT","name":"Baidu: ERNIE 4.5 VL 424B A47B ","created":1751300903,"description":"ERNIE-4.5-VL-424B-A47B is a multimodal Mixture-of-Experts (MoE) model from Baidu’s ERNIE 4.5 series, featuring 424B total parameters with 47B active per token. It is trained jointly on text and image data...","context_length":123000,"architecture":{"modality":"text+image->text","input_modalities":["image","text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000042","completion":"0.00000125"},"top_provider":{"context_length":123000,"max_completion_tokens":16000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","repetition_penalty","seed","stop","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/baidu/ernie-4.5-vl-424b-a47b/endpoints"},"reasoning":{"mandatory":false}},{"id":"mistralai/mistral-small-3.2-24b-instruct","canonical_slug":"mistralai/mistral-small-3.2-24b-instruct-2506","hugging_face_id":"mistralai/Mistral-Small-3.2-24B-Instruct-2506","name":"Mistral: Mistral Small 3.2 24B","created":1750443016,"description":"Mistral-Small-3.2-24B-Instruct-2506 is an updated 24B parameter model from Mistral optimized for instruction following, repetition reduction, and improved function calling. Compared to the 3.1 release, version 3.2 significantly improves accuracy on...","context_length":256000,"architecture":{"modality":"text+image->text","input_modalities":["image","text"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.000000075","completion":"0.0000002"},"top_provider":{"context_length":128000,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-small-3.2-24b-instruct-2506/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"codecategories","elo":922,"win_rate":39.8,"rank":120},{"arena":"models","category":"dataviz","elo":943,"win_rate":43.3,"rank":113},{"arena":"models","category":"gamedev","elo":911,"win_rate":39.4,"rank":121},{"arena":"models","category":"uicomponent","elo":922,"win_rate":40.5,"rank":114},{"arena":"models","category":"website","elo":908,"win_rate":38.3,"rank":123}]}},{"id":"minimax/minimax-m1","canonical_slug":"minimax/minimax-m1","hugging_face_id":"","name":"MiniMax: MiniMax M1","created":1750200414,"description":"MiniMax-M1 is a large-scale, open-weight reasoning model designed for extended context and high-efficiency inference. It leverages a hybrid Mixture-of-Experts (MoE) architecture paired with a custom \"lightning attention\" mechanism, allowing it...","context_length":1000000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000055","completion":"0.0000022"},"top_provider":{"context_length":1000000,"max_completion_tokens":40000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","repetition_penalty","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/minimax/minimax-m1/endpoints"},"reasoning":{"mandatory":false}},{"id":"google/gemini-2.5-flash","canonical_slug":"google/gemini-2.5-flash","hugging_face_id":"","name":"Google: Gemini 2.5 Flash","created":1750172488,"description":"Gemini 2.5 Flash is Google's state-of-the-art workhorse model, specifically designed for advanced reasoning, coding, mathematics, and scientific tasks. It includes built-in \"thinking\" capabilities, enabling it to provide responses with greater...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["file","image","text","audio","video"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.0000003","completion":"0.0000025","image":"0.0000003","audio":"0.000001","input_audio_cache":"0.0000001","web_search":"0.014","internal_reasoning":"0.0000025","input_cache_read":"0.00000003","input_cache_write":"0.0000000833333333333333"},"top_provider":{"context_length":1048576,"max_completion_tokens":65535,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-2.5-flash/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1099,"win_rate":47.4,"rank":89},{"arena":"models","category":"codecategories","elo":1119,"win_rate":46.9,"rank":90},{"arena":"models","category":"dataviz","elo":1150,"win_rate":49.1,"rank":77},{"arena":"models","category":"gamedev","elo":1089,"win_rate":44.3,"rank":93},{"arena":"models","category":"uicomponent","elo":1108,"win_rate":48.9,"rank":85},{"arena":"models","category":"website","elo":1127,"win_rate":47.1,"rank":90},{"arena":"models","category":"svg","elo":1045,"win_rate":43.1,"rank":74}]},"reasoning":{"mandatory":false}},{"id":"google/gemini-2.5-flash:batch","canonical_slug":"google/gemini-2.5-flash","hugging_face_id":"","name":"Google: Gemini 2.5 Flash (batch)","created":1750172488,"description":"Gemini 2.5 Flash is Google's state-of-the-art workhorse model, specifically designed for advanced reasoning, coding, mathematics, and scientific tasks. It includes built-in \"thinking\" capabilities, enabling it to provide responses with greater...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["file","image","text","audio","video"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.00000125","image":"0.00000015","audio":"0.0000005","input_audio_cache":"0.0000001","web_search":"0.014","internal_reasoning":"0.00000125","input_cache_read":"0.00000003"},"top_provider":{"context_length":1048576,"max_completion_tokens":65535,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-2.5-flash/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1099,"win_rate":47.4,"rank":89},{"arena":"models","category":"codecategories","elo":1119,"win_rate":46.9,"rank":90},{"arena":"models","category":"dataviz","elo":1150,"win_rate":49.1,"rank":77},{"arena":"models","category":"gamedev","elo":1089,"win_rate":44.3,"rank":93},{"arena":"models","category":"uicomponent","elo":1108,"win_rate":48.9,"rank":85},{"arena":"models","category":"website","elo":1127,"win_rate":47.1,"rank":90},{"arena":"models","category":"svg","elo":1045,"win_rate":43.1,"rank":74}]},"reasoning":{"mandatory":false}},{"id":"google/gemini-2.5-pro","canonical_slug":"google/gemini-2.5-pro","hugging_face_id":"","name":"Google: Gemini 2.5 Pro","created":1750169544,"description":"Gemini 2.5 Pro is Google’s state-of-the-art AI model designed for advanced reasoning, coding, mathematics, and scientific tasks. It employs “thinking” capabilities, enabling it to reason through responses with enhanced accuracy...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","file","audio","video"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.00001","image":"0.00000125","audio":"0.00000125","input_audio_cache":"0.000000125","web_search":"0.014","internal_reasoning":"0.00001","input_cache_read":"0.000000125","input_cache_write":"0.000000375","overrides":[{"min_prompt_tokens":200000,"prompt":"0.0000025","completion":"0.000015","audio":"0.0000025","input_audio_cache":"0.00000025","input_cache_read":"0.00000025"}]},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-2.5-pro/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1109,"win_rate":50.6,"rank":84},{"arena":"models","category":"codecategories","elo":1167,"win_rate":57.5,"rank":70},{"arena":"models","category":"dataviz","elo":1237,"win_rate":68.2,"rank":39},{"arena":"models","category":"gamedev","elo":1134,"win_rate":54.2,"rank":76},{"arena":"models","category":"uicomponent","elo":1155,"win_rate":57.5,"rank":70},{"arena":"models","category":"website","elo":1179,"win_rate":58.4,"rank":69}],"artificial_analysis":{"intelligence_index":16.7,"coding_index":33.3,"agentic_index":3.5}},"reasoning":{"mandatory":true}},{"id":"google/gemini-2.5-pro:batch","canonical_slug":"google/gemini-2.5-pro","hugging_face_id":"","name":"Google: Gemini 2.5 Pro (batch)","created":1750169544,"description":"Gemini 2.5 Pro is Google’s state-of-the-art AI model designed for advanced reasoning, coding, mathematics, and scientific tasks. It employs “thinking” capabilities, enabling it to reason through responses with enhanced accuracy...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","file","audio","video"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.000000625","completion":"0.000005","image":"0.000000625","audio":"0.000000625","input_audio_cache":"0.000000125","web_search":"0.014","internal_reasoning":"0.000005","input_cache_read":"0.000000125","overrides":[{"min_prompt_tokens":200000,"prompt":"0.00000125","completion":"0.0000075","audio":"0.00000125","input_audio_cache":"0.00000025","input_cache_read":"0.00000025"}]},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-2.5-pro/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1109,"win_rate":50.6,"rank":84},{"arena":"models","category":"codecategories","elo":1167,"win_rate":57.5,"rank":70},{"arena":"models","category":"dataviz","elo":1237,"win_rate":68.2,"rank":39},{"arena":"models","category":"gamedev","elo":1134,"win_rate":54.2,"rank":76},{"arena":"models","category":"uicomponent","elo":1155,"win_rate":57.5,"rank":70},{"arena":"models","category":"website","elo":1179,"win_rate":58.4,"rank":69}],"artificial_analysis":{"intelligence_index":16.7,"coding_index":33.3,"agentic_index":3.5}},"reasoning":{"mandatory":true}},{"id":"openai/o3-pro","canonical_slug":"openai/o3-pro-2025-06-10","hugging_face_id":"","name":"OpenAI: o3 Pro","created":1749598352,"description":"The o-series of models are trained with reinforcement learning to think before they answer and perform complex reasoning. The o3-pro model uses more compute to think harder and provide consistently...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","file","image"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00002","completion":"0.00008","web_search":"0.01"},"top_provider":{"context_length":200000,"max_completion_tokens":100000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/o3-pro-2025-06-10/endpoints"},"reasoning":{"mandatory":false}},{"id":"google/gemini-2.5-pro-preview","canonical_slug":"google/gemini-2.5-pro-preview-06-05","hugging_face_id":"","name":"Google: Gemini 2.5 Pro Preview 06-05","created":1749137257,"description":"Gemini 2.5 Pro is Google’s state-of-the-art AI model designed for advanced reasoning, coding, mathematics, and scientific tasks. It employs “thinking” capabilities, enabling it to reason through responses with enhanced accuracy...","context_length":1048576,"architecture":{"modality":"text+image+file+audio->text","input_modalities":["file","image","text","audio"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.00001","image":"0.00000125","audio":"0.00000125","input_audio_cache":"0.000000125","web_search":"0.014","internal_reasoning":"0.00001","input_cache_read":"0.000000125","input_cache_write":"0.000000375","overrides":[{"min_prompt_tokens":200000,"prompt":"0.0000025","completion":"0.000015","audio":"0.0000025","input_audio_cache":"0.00000025","input_cache_read":"0.00000025"}]},"top_provider":{"context_length":1048576,"max_completion_tokens":65536,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-2.5-pro-preview-06-05/endpoints"},"reasoning":{"mandatory":true}},{"id":"deepseek/deepseek-r1-0528","canonical_slug":"deepseek/deepseek-r1-0528","hugging_face_id":"deepseek-ai/DeepSeek-R1-0528","name":"DeepSeek: R1 0528","created":1748455170,"description":"May 28th update to the [original DeepSeek R1](/deepseek/deepseek-r1) Performance on par with [OpenAI o1](/openai/o1), but open-sourced and with fully open reasoning tokens. It's 671B parameters in size, with 37B active...","context_length":163840,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":"deepseek-r1"},"pricing":{"prompt":"0.0000005","completion":"0.00000215","input_cache_read":"0.00000035"},"top_provider":{"context_length":163840,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-r1-0528/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1139,"win_rate":53.3,"rank":71},{"arena":"models","category":"codecategories","elo":1153,"win_rate":52.6,"rank":78},{"arena":"models","category":"dataviz","elo":1198,"win_rate":60.9,"rank":53},{"arena":"models","category":"gamedev","elo":1120,"win_rate":49.3,"rank":82},{"arena":"models","category":"svg","elo":1063,"win_rate":48.7,"rank":68},{"arena":"models","category":"uicomponent","elo":1118,"win_rate":54.9,"rank":80},{"arena":"models","category":"website","elo":1161,"win_rate":52.7,"rank":75}]},"reasoning":{"mandatory":true}},{"id":"anthropic/claude-opus-4","canonical_slug":"anthropic/claude-4-opus-20250522","hugging_face_id":"","name":"Anthropic: Claude Opus 4","created":1747931245,"description":"Claude Opus 4 is benchmarked as the world’s best coding model, at time of release, bringing sustained performance on complex, long-running tasks and agent workflows. It sets new benchmarks in...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000015","completion":"0.000075","web_search":"0.01","input_cache_read":"0.0000015","input_cache_write":"0.00001875","input_cache_write_1h":"0.00003"},"top_provider":{"context_length":200000,"max_completion_tokens":32000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","stop","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4-opus-20250522/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1167,"win_rate":57.7,"rank":60},{"arena":"models","category":"codecategories","elo":1176,"win_rate":55.6,"rank":69},{"arena":"models","category":"dataviz","elo":1161,"win_rate":57.9,"rank":71},{"arena":"models","category":"gamedev","elo":1193,"win_rate":59.9,"rank":55},{"arena":"models","category":"svg","elo":1152,"win_rate":57.7,"rank":49},{"arena":"models","category":"uicomponent","elo":1168,"win_rate":59.2,"rank":65},{"arena":"models","category":"website","elo":1177,"win_rate":54.6,"rank":70}]},"reasoning":{"mandatory":false}},{"id":"anthropic/claude-sonnet-4","canonical_slug":"anthropic/claude-4-sonnet-20250522","hugging_face_id":"","name":"Anthropic: Claude Sonnet 4","created":1747930371,"description":"Claude Sonnet 4 significantly enhances the capabilities of its predecessor, Sonnet 3.7, excelling in both coding and reasoning tasks with improved precision and controllability. Achieving state-of-the-art performance on SWE-bench (72.7%),...","context_length":1000000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.000003","completion":"0.000015","web_search":"0.01","input_cache_read":"0.0000003","input_cache_write":"0.00000375","input_cache_write_1h":"0.000006","overrides":[{"min_prompt_tokens":200000,"prompt":"0.000006","completion":"0.0000225","input_cache_read":"0.0000006","input_cache_write":"0.0000075","input_cache_write_1h":"0.000012"}]},"top_provider":{"context_length":200000,"max_completion_tokens":64000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-4-sonnet-20250522/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1168,"win_rate":57.8,"rank":59},{"arena":"models","category":"codecategories","elo":1156,"win_rate":53.4,"rank":75},{"arena":"models","category":"dataviz","elo":1173,"win_rate":56.5,"rank":69},{"arena":"models","category":"gamedev","elo":1161,"win_rate":54.8,"rank":65},{"arena":"models","category":"svg","elo":1103,"win_rate":51.1,"rank":59},{"arena":"models","category":"uicomponent","elo":1144,"win_rate":57.9,"rank":74},{"arena":"models","category":"website","elo":1158,"win_rate":52.4,"rank":78}],"artificial_analysis":{"intelligence_index":null,"coding_index":37.6,"agentic_index":null}},"reasoning":{"mandatory":false}},{"id":"mistralai/mistral-medium-3","canonical_slug":"mistralai/mistral-medium-3","hugging_face_id":"","name":"Mistral: Mistral Medium 3","created":1746627341,"description":"Mistral Medium 3 is a high-performance enterprise-grade language model designed to deliver frontier-level capabilities at significantly reduced operational cost. It balances state-of-the-art reasoning and multimodal performance with 8× lower cost...","context_length":131072,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.0000004","completion":"0.000002","input_cache_read":"0.00000004"},"top_provider":{"context_length":131072,"max_completion_tokens":104857,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-medium-3/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1112,"win_rate":54.7,"rank":80},{"arena":"models","category":"codecategories","elo":1084,"win_rate":48.1,"rank":97},{"arena":"models","category":"dataviz","elo":1049,"win_rate":46,"rank":100},{"arena":"models","category":"gamedev","elo":1041,"win_rate":45.2,"rank":103},{"arena":"models","category":"uicomponent","elo":1040,"win_rate":49.6,"rank":99},{"arena":"models","category":"website","elo":1091,"win_rate":47.6,"rank":100}]}},{"id":"google/gemini-2.5-pro-preview-05-06","canonical_slug":"google/gemini-2.5-pro-preview-03-25","hugging_face_id":"","name":"Google: Gemini 2.5 Pro Preview 05-06","created":1746578513,"description":"Gemini 2.5 Pro is Google’s state-of-the-art AI model designed for advanced reasoning, coding, mathematics, and scientific tasks. It employs “thinking” capabilities, enabling it to reason through responses with enhanced accuracy...","context_length":1048576,"architecture":{"modality":"text+image+file+audio+video->text","input_modalities":["text","image","file","audio","video"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.00001","image":"0.00000125","audio":"0.00000125","input_audio_cache":"0.000000125","web_search":"0.014","internal_reasoning":"0.00001","input_cache_read":"0.000000125","input_cache_write":"0.000000375","overrides":[{"min_prompt_tokens":200000,"prompt":"0.0000025","completion":"0.000015","audio":"0.0000025","input_audio_cache":"0.00000025","input_cache_read":"0.00000025"}]},"top_provider":{"context_length":1048576,"max_completion_tokens":65535,"is_moderated":false},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemini-2.5-pro-preview-03-25/endpoints"},"reasoning":{"mandatory":true}},{"id":"meta-llama/llama-guard-4-12b","canonical_slug":"meta-llama/llama-guard-4-12b","hugging_face_id":"meta-llama/Llama-Guard-4-12B","name":"Meta: Llama Guard 4 12B","created":1745975193,"description":"Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification. Similar to previous versions, it can be used to classify content in both LLM...","context_length":163840,"architecture":{"modality":"text+image->text","input_modalities":["image","text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000018","completion":"0.00000018"},"top_provider":{"context_length":163840,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-08-31","expiration_date":null,"links":{"details":"/api/v1/models/meta-llama/llama-guard-4-12b/endpoints"}},{"id":"qwen/qwen3-30b-a3b","canonical_slug":"qwen/qwen3-30b-a3b-04-28","hugging_face_id":"Qwen/Qwen3-30B-A3B","name":"Qwen: Qwen3 30B A3B","created":1745878604,"description":"Qwen3, the latest generation in the Qwen large language model series, features both dense and mixture-of-experts (MoE) architectures to excel in reasoning, multilingual support, and advanced agent tasks. Its unique...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":"qwen3"},"pricing":{"prompt":"0.00000012","completion":"0.0000005"},"top_provider":{"context_length":40960,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-30b-a3b-04-28/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"codecategories","elo":957,"win_rate":37.5,"rank":119},{"arena":"models","category":"dataviz","elo":981,"win_rate":39,"rank":109},{"arena":"models","category":"gamedev","elo":921,"win_rate":33.9,"rank":120},{"arena":"models","category":"uicomponent","elo":961,"win_rate":42.4,"rank":109},{"arena":"models","category":"website","elo":967,"win_rate":37.7,"rank":121}]},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"qwen/qwen3-8b","canonical_slug":"qwen/qwen3-8b-04-28","hugging_face_id":"Qwen/Qwen3-8B","name":"Qwen: Qwen3 8B","created":1745876632,"description":"Qwen3-8B is a dense 8.2B parameter causal language model from the Qwen3 series, designed for both reasoning-heavy tasks and efficient dialogue. It supports seamless switching between \"thinking\" mode for math,...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":"qwen3"},"pricing":{"prompt":"0.000000117","completion":"0.000000455"},"top_provider":{"context_length":131072,"max_completion_tokens":8192,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","response_format","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":0.6,"top_p":0.95,"top_k":20,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-8b-04-28/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":5.2,"coding_index":9,"agentic_index":0.8}},"reasoning":{"mandatory":false,"default_enabled":true}},{"id":"qwen/qwen3-14b","canonical_slug":"qwen/qwen3-14b-04-28","hugging_face_id":"Qwen/Qwen3-14B","name":"Qwen: Qwen3 14B","created":1745876478,"description":"Qwen3-14B is a dense 14.8B parameter causal language model from the Qwen3 series, designed for both complex reasoning and efficient dialogue. It supports seamless switching between a \"thinking\" mode for...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":"qwen3"},"pricing":{"prompt":"0.00000012","completion":"0.00000024"},"top_provider":{"context_length":40960,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-14b-04-28/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":6.4,"coding_index":13.8,"agentic_index":0.9}},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3-32b","canonical_slug":"qwen/qwen3-32b-04-28","hugging_face_id":"Qwen/Qwen3-32B","name":"Qwen: Qwen3 32B","created":1745875945,"description":"Qwen3-32B is a dense 32.8B parameter causal language model from the Qwen3 series, optimized for both complex reasoning and efficient dialogue. It supports seamless switching between a \"thinking\" mode for...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":"qwen3"},"pricing":{"prompt":"0.00000008","completion":"0.00000028"},"top_provider":{"context_length":40960,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","max_tokens","min_p","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-32b-04-28/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":7.2,"coding_index":15.3,"agentic_index":0.9}},"reasoning":{"mandatory":false}},{"id":"qwen/qwen3-235b-a22b","canonical_slug":"qwen/qwen3-235b-a22b-04-28","hugging_face_id":"Qwen/Qwen3-235B-A22B","name":"Qwen: Qwen3 235B A22B","created":1745875757,"description":"Qwen3-235B-A22B is a 235B parameter mixture-of-experts (MoE) model developed by Qwen, activating 22B parameters per forward pass. It supports seamless switching between a \"thinking\" mode for complex reasoning, math, and...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen3","instruct_type":"qwen3"},"pricing":{"prompt":"0.000000455","completion":"0.00000182"},"top_provider":{"context_length":131072,"max_completion_tokens":8192,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","response_format","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen3-235b-a22b-04-28/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":887,"win_rate":24.5,"rank":112},{"arena":"models","category":"codecategories","elo":1017,"win_rate":38.2,"rank":112},{"arena":"models","category":"dataviz","elo":1011,"win_rate":40,"rank":105},{"arena":"models","category":"gamedev","elo":950,"win_rate":32.9,"rank":117},{"arena":"models","category":"uicomponent","elo":974,"win_rate":38.6,"rank":108},{"arena":"models","category":"website","elo":1043,"win_rate":40.4,"rank":109}]},"reasoning":{"mandatory":false}},{"id":"openai/o4-mini-high","canonical_slug":"openai/o4-mini-high-2025-04-16","hugging_face_id":"","name":"OpenAI: o4 Mini High","created":1744824212,"description":"OpenAI o4-mini-high is the same model as [o4-mini](/openai/o4-mini) with reasoning_effort set to high. OpenAI o4-mini is a compact reasoning model in the o-series, optimized for fast, cost-efficient performance while retaining...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000011","completion":"0.0000044","web_search":"0.01","input_cache_read":"0.000000275"},"top_provider":{"context_length":200000,"max_completion_tokens":100000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/o4-mini-high-2025-04-16/endpoints"},"reasoning":{"mandatory":true,"supported_efforts":["high"],"default_effort":"high"}},{"id":"openai/o3","canonical_slug":"openai/o3-2025-04-16","hugging_face_id":"","name":"OpenAI: o3","created":1744823457,"description":"o3 is a well-rounded and powerful model across domains. It sets a new standard for math, science, coding, and visual reasoning tasks. It also excels at technical writing and instruction-following....","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000008","web_search":"0.01","input_cache_read":"0.0000005"},"top_provider":{"context_length":200000,"max_completion_tokens":100000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/o3-2025-04-16/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"codecategories","elo":1035,"win_rate":51.9,"rank":107},{"arena":"models","category":"dataviz","elo":1200,"win_rate":48.1,"rank":52},{"arena":"models","category":"gamedev","elo":1057,"win_rate":56.9,"rank":99},{"arena":"models","category":"uicomponent","elo":1032,"win_rate":53.3,"rank":102},{"arena":"models","category":"website","elo":1048,"win_rate":53.8,"rank":108}]},"reasoning":{"mandatory":false}},{"id":"openai/o3:batch","canonical_slug":"openai/o3-2025-04-16","hugging_face_id":"","name":"OpenAI: o3 (batch)","created":1744823457,"description":"o3 is a well-rounded and powerful model across domains. It sets a new standard for math, science, coding, and visual reasoning tasks. It also excels at technical writing and instruction-following....","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000004","web_search":"0.01","input_cache_read":"0.00000025"},"top_provider":{"context_length":200000,"max_completion_tokens":100000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/o3-2025-04-16/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"codecategories","elo":1035,"win_rate":51.9,"rank":107},{"arena":"models","category":"dataviz","elo":1200,"win_rate":48.1,"rank":52},{"arena":"models","category":"gamedev","elo":1057,"win_rate":56.9,"rank":99},{"arena":"models","category":"uicomponent","elo":1032,"win_rate":53.3,"rank":102},{"arena":"models","category":"website","elo":1048,"win_rate":53.8,"rank":108}]},"reasoning":{"mandatory":false}},{"id":"openai/o4-mini","canonical_slug":"openai/o4-mini-2025-04-16","hugging_face_id":"","name":"OpenAI: o4 Mini","created":1744820942,"description":"OpenAI o4-mini is a compact reasoning model in the o-series, optimized for fast, cost-efficient performance while retaining strong multimodal and agentic capabilities. It supports tool use and demonstrates competitive reasoning...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000011","completion":"0.0000044","web_search":"0.01","input_cache_read":"0.000000275"},"top_provider":{"context_length":200000,"max_completion_tokens":100000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/o4-mini-2025-04-16/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":882,"win_rate":34,"rank":113},{"arena":"models","category":"codecategories","elo":991,"win_rate":46.4,"rank":114},{"arena":"models","category":"dataviz","elo":1006,"win_rate":50,"rank":106},{"arena":"models","category":"gamedev","elo":1026,"win_rate":50,"rank":105},{"arena":"models","category":"uicomponent","elo":999,"win_rate":46.9,"rank":104},{"arena":"models","category":"website","elo":997,"win_rate":47.1,"rank":116}]},"reasoning":{"mandatory":false}},{"id":"openai/o4-mini:batch","canonical_slug":"openai/o4-mini-2025-04-16","hugging_face_id":"","name":"OpenAI: o4 Mini (batch)","created":1744820942,"description":"OpenAI o4-mini is a compact reasoning model in the o-series, optimized for fast, cost-efficient performance while retaining strong multimodal and agentic capabilities. It supports tool use and demonstrates competitive reasoning...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000055","completion":"0.0000022","web_search":"0.01","input_cache_read":"0.0000001375"},"top_provider":{"context_length":200000,"max_completion_tokens":100000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/o4-mini-2025-04-16/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":882,"win_rate":34,"rank":113},{"arena":"models","category":"codecategories","elo":991,"win_rate":46.4,"rank":114},{"arena":"models","category":"dataviz","elo":1006,"win_rate":50,"rank":106},{"arena":"models","category":"gamedev","elo":1026,"win_rate":50,"rank":105},{"arena":"models","category":"uicomponent","elo":999,"win_rate":46.9,"rank":104},{"arena":"models","category":"website","elo":997,"win_rate":47.1,"rank":116}]},"reasoning":{"mandatory":false}},{"id":"openai/gpt-4.1","canonical_slug":"openai/gpt-4.1-2025-04-14","hugging_face_id":"","name":"OpenAI: GPT-4.1","created":1744651385,"description":"GPT-4.1 is a flagship large language model optimized for advanced instruction following, real-world software engineering, and long-context reasoning. It supports a 1 million token context window and outperforms GPT-4o and...","context_length":1047576,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000008","web_search":"0.01","input_cache_read":"0.0000005"},"top_provider":{"context_length":1047576,"max_completion_tokens":32768,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_completion_tokens","max_tokens","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4.1-2025-04-14/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":878,"win_rate":30.9,"rank":114},{"arena":"models","category":"codecategories","elo":1040,"win_rate":50.9,"rank":106},{"arena":"models","category":"dataviz","elo":1118,"win_rate":59.5,"rank":86},{"arena":"models","category":"gamedev","elo":1102,"win_rate":59.1,"rank":88},{"arena":"models","category":"uicomponent","elo":1015,"win_rate":49.7,"rank":103},{"arena":"models","category":"website","elo":1051,"win_rate":52.3,"rank":107}]}},{"id":"openai/gpt-4.1:batch","canonical_slug":"openai/gpt-4.1-2025-04-14","hugging_face_id":"","name":"OpenAI: GPT-4.1 (batch)","created":1744651385,"description":"GPT-4.1 is a flagship large language model optimized for advanced instruction following, real-world software engineering, and long-context reasoning. It supports a 1 million token context window and outperforms GPT-4o and...","context_length":1047576,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000004","web_search":"0.01","input_cache_read":"0.00000025"},"top_provider":{"context_length":1047576,"max_completion_tokens":32768,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_tokens","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4.1-2025-04-14/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":878,"win_rate":30.9,"rank":114},{"arena":"models","category":"codecategories","elo":1040,"win_rate":50.9,"rank":106},{"arena":"models","category":"dataviz","elo":1118,"win_rate":59.5,"rank":86},{"arena":"models","category":"gamedev","elo":1102,"win_rate":59.1,"rank":88},{"arena":"models","category":"uicomponent","elo":1015,"win_rate":49.7,"rank":103},{"arena":"models","category":"website","elo":1051,"win_rate":52.3,"rank":107}]}},{"id":"openai/gpt-4.1-mini","canonical_slug":"openai/gpt-4.1-mini-2025-04-14","hugging_face_id":"","name":"OpenAI: GPT-4.1 Mini","created":1744651381,"description":"GPT-4.1 Mini is a mid-sized model delivering performance competitive with GPT-4o at substantially lower latency and cost. It retains a 1 million token context window and scores 45.1% on hard...","context_length":1047576,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000004","completion":"0.0000016","web_search":"0.01","input_cache_read":"0.0000001"},"top_provider":{"context_length":1047576,"max_completion_tokens":32768,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_completion_tokens","max_tokens","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4.1-mini-2025-04-14/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":865,"win_rate":30.5,"rank":115},{"arena":"models","category":"codecategories","elo":1008,"win_rate":47.5,"rank":113},{"arena":"models","category":"dataviz","elo":1048,"win_rate":49.2,"rank":101},{"arena":"models","category":"gamedev","elo":1093,"win_rate":58.5,"rank":91},{"arena":"models","category":"uicomponent","elo":976,"win_rate":45.4,"rank":107},{"arena":"models","category":"website","elo":1010,"win_rate":47.8,"rank":115}],"artificial_analysis":{"intelligence_index":null,"coding_index":20.2,"agentic_index":null}}},{"id":"openai/gpt-4.1-mini:batch","canonical_slug":"openai/gpt-4.1-mini-2025-04-14","hugging_face_id":"","name":"OpenAI: GPT-4.1 Mini (batch)","created":1744651381,"description":"GPT-4.1 Mini is a mid-sized model delivering performance competitive with GPT-4o at substantially lower latency and cost. It retains a 1 million token context window and scores 45.1% on hard...","context_length":1047576,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.0000008","web_search":"0.01","input_cache_read":"0.00000005"},"top_provider":{"context_length":1047576,"max_completion_tokens":32768,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_tokens","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4.1-mini-2025-04-14/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":865,"win_rate":30.5,"rank":115},{"arena":"models","category":"codecategories","elo":1008,"win_rate":47.5,"rank":113},{"arena":"models","category":"dataviz","elo":1048,"win_rate":49.2,"rank":101},{"arena":"models","category":"gamedev","elo":1093,"win_rate":58.5,"rank":91},{"arena":"models","category":"uicomponent","elo":976,"win_rate":45.4,"rank":107},{"arena":"models","category":"website","elo":1010,"win_rate":47.8,"rank":115}],"artificial_analysis":{"intelligence_index":null,"coding_index":20.2,"agentic_index":null}}},{"id":"openai/gpt-4.1-nano","canonical_slug":"openai/gpt-4.1-nano-2025-04-14","hugging_face_id":"","name":"OpenAI: GPT-4.1 Nano","created":1744651369,"description":"For tasks that demand low latency, GPT‑4.1 nano is the fastest and cheapest model in the GPT-4.1 series. It delivers exceptional performance at a small size with its 1 million...","context_length":1047576,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000004","web_search":"0.01","input_cache_read":"0.000000025"},"top_provider":{"context_length":1047576,"max_completion_tokens":32768,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_completion_tokens","max_tokens","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4.1-nano-2025-04-14/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":953,"win_rate":46,"rank":108},{"arena":"models","category":"codecategories","elo":978,"win_rate":47.3,"rank":116},{"arena":"models","category":"dataviz","elo":906,"win_rate":41.1,"rank":118},{"arena":"models","category":"gamedev","elo":993,"win_rate":49.6,"rank":111},{"arena":"models","category":"uicomponent","elo":931,"win_rate":43.9,"rank":113},{"arena":"models","category":"website","elo":985,"win_rate":48.1,"rank":117}],"artificial_analysis":{"intelligence_index":null,"coding_index":11.1,"agentic_index":null}}},{"id":"openai/gpt-4.1-nano:batch","canonical_slug":"openai/gpt-4.1-nano-2025-04-14","hugging_face_id":"","name":"OpenAI: GPT-4.1 Nano (batch)","created":1744651369,"description":"For tasks that demand low latency, GPT‑4.1 nano is the fastest and cheapest model in the GPT-4.1 series. It delivers exceptional performance at a small size with its 1 million...","context_length":1047576,"architecture":{"modality":"text+image+file->text","input_modalities":["image","text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000005","completion":"0.0000002","web_search":"0.01","input_cache_read":"0.0000000125"},"top_provider":{"context_length":1047576,"max_completion_tokens":32768,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_tokens","response_format","seed","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4.1-nano-2025-04-14/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":953,"win_rate":46,"rank":108},{"arena":"models","category":"codecategories","elo":978,"win_rate":47.3,"rank":116},{"arena":"models","category":"dataviz","elo":906,"win_rate":41.1,"rank":118},{"arena":"models","category":"gamedev","elo":993,"win_rate":49.6,"rank":111},{"arena":"models","category":"uicomponent","elo":931,"win_rate":43.9,"rank":113},{"arena":"models","category":"website","elo":985,"win_rate":48.1,"rank":117}],"artificial_analysis":{"intelligence_index":null,"coding_index":11.1,"agentic_index":null}}},{"id":"meta-llama/llama-4-maverick","canonical_slug":"meta-llama/llama-4-maverick-17b-128e-instruct","hugging_face_id":"meta-llama/Llama-4-Maverick-17B-128E-Instruct","name":"Meta: Llama 4 Maverick","created":1743881822,"description":"Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...","context_length":1048576,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Llama4","instruct_type":null},"pricing":{"prompt":"0.0000001875","completion":"0.0000006525"},"top_provider":{"context_length":128000,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-08-31","expiration_date":null,"links":{"details":"/api/v1/models/meta-llama/llama-4-maverick-17b-128e-instruct/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":928,"win_rate":40.2,"rank":110},{"arena":"models","category":"codecategories","elo":894,"win_rate":35.8,"rank":122},{"arena":"models","category":"dataviz","elo":896,"win_rate":38.4,"rank":119},{"arena":"models","category":"gamedev","elo":860,"win_rate":33.7,"rank":122},{"arena":"models","category":"uicomponent","elo":914,"win_rate":40.8,"rank":115},{"arena":"models","category":"website","elo":883,"win_rate":34.4,"rank":125}],"artificial_analysis":{"intelligence_index":9.3,"coding_index":16.3,"agentic_index":0.6}}},{"id":"meta-llama/llama-4-scout","canonical_slug":"meta-llama/llama-4-scout-17b-16e-instruct","hugging_face_id":"meta-llama/Llama-4-Scout-17B-16E-Instruct","name":"Meta: Llama 4 Scout","created":1743881519,"description":"Llama 4 Scout 17B Instruct (16E) is a mixture-of-experts (MoE) language model developed by Meta, activating 17 billion parameters out of a total of 109B. It supports native multimodal input...","context_length":1310720,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Llama4","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000003"},"top_provider":{"context_length":327680,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-08-31","expiration_date":null,"links":{"details":"/api/v1/models/meta-llama/llama-4-scout-17b-16e-instruct/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"codecategories","elo":804,"win_rate":26.6,"rank":125},{"arena":"models","category":"dataviz","elo":910,"win_rate":39.3,"rank":117},{"arena":"models","category":"gamedev","elo":796,"win_rate":27.4,"rank":124},{"arena":"models","category":"uicomponent","elo":783,"win_rate":25.5,"rank":120},{"arena":"models","category":"website","elo":762,"win_rate":22.7,"rank":131}],"artificial_analysis":{"intelligence_index":6.5,"coding_index":8.2,"agentic_index":0.5}}},{"id":"deepseek/deepseek-chat-v3-0324","canonical_slug":"deepseek/deepseek-chat-v3-0324","hugging_face_id":"deepseek-ai/DeepSeek-V3-0324","name":"DeepSeek: DeepSeek V3 0324","created":1742824755,"description":"DeepSeek V3, a 685B-parameter, mixture-of-experts model, is the latest iteration of the flagship chat model family from the DeepSeek team. It succeeds the [DeepSeek V3](/deepseek/deepseek-chat-v3) model and performs really well...","context_length":163840,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.000001"},"top_provider":{"context_length":163840,"max_completion_tokens":147456,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-07-31","expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-chat-v3-0324/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":9.7,"coding_index":21.2,"agentic_index":0.8}}},{"id":"openai/o1-pro","canonical_slug":"openai/o1-pro","hugging_face_id":"","name":"OpenAI: o1-pro","created":1742423211,"description":"The o1 series of models are trained with reinforcement learning to think before they answer and perform complex reasoning. The o1-pro model uses more compute to think harder and provide...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00015","completion":"0.0006","web_search":"0.01"},"top_provider":{"context_length":200000,"max_completion_tokens":100000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","structured_outputs"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/o1-pro/endpoints"},"reasoning":{"mandatory":false}},{"id":"mistralai/mistral-small-3.1-24b-instruct","canonical_slug":"mistralai/mistral-small-3.1-24b-instruct-2503","hugging_face_id":"mistralai/Mistral-Small-3.1-24B-Instruct-2503","name":"Mistral: Mistral Small 3.1 24B","created":1742238937,"description":"Mistral Small 3.1 24B Instruct is an upgraded variant of Mistral Small 3 (2501), featuring 24 billion parameters with advanced multimodal capabilities. It provides state-of-the-art performance in text-based reasoning and...","context_length":128000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.000000351","completion":"0.000000555"},"top_provider":{"context_length":128000,"max_completion_tokens":102400,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","seed","stop","temperature","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-small-3.1-24b-instruct-2503/endpoints"}},{"id":"google/gemma-3-4b-it","canonical_slug":"google/gemma-3-4b-it","hugging_face_id":"google/gemma-3-4b-it","name":"Google: Gemma 3 4B","created":1741905510,"description":"Gemma 3 introduces multimodality, supporting vision-language input and text outputs. It handles context windows up to 128k tokens, understands over 140 languages, and offers improved math, reasoning, and chat capabilities,...","context_length":131072,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":"gemma"},"pricing":{"prompt":"0.00000005","completion":"0.0000001"},"top_provider":{"context_length":131072,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-08-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemma-3-4b-it/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":2.7,"agentic_index":null}}},{"id":"google/gemma-3-12b-it","canonical_slug":"google/gemma-3-12b-it","hugging_face_id":"google/gemma-3-12b-it","name":"Google: Gemma 3 12B","created":1741902625,"description":"Gemma 3 introduces multimodality, supporting vision-language input and text outputs. It handles context windows up to 128k tokens, understands over 140 languages, and offers improved math, reasoning, and chat capabilities,...","context_length":131072,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":"gemma"},"pricing":{"prompt":"0.00000005","completion":"0.00000015"},"top_provider":{"context_length":131072,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-08-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemma-3-12b-it/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":3.8,"coding_index":5.8,"agentic_index":0.1}}},{"id":"cohere/command-a","canonical_slug":"cohere/command-a-03-2025","hugging_face_id":"CohereForAI/c4ai-command-a-03-2025","name":"Cohere: Command A","created":1741894342,"description":"Command A is an open-weights 111B parameter model with a 256k context window focused on delivering great performance across agentic, multilingual, and coding use cases. Compared to other leading proprietary...","context_length":256000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.00001"},"top_provider":{"context_length":256000,"max_completion_tokens":8192,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-08-31","expiration_date":null,"links":{"details":"/api/v1/models/cohere/command-a-03-2025/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":13.9,"coding_index":27.8,"agentic_index":3.6}}},{"id":"rekaai/reka-flash-3","canonical_slug":"rekaai/reka-flash-3","hugging_face_id":"RekaAI/reka-flash-3","name":"Reka Flash 3","created":1741812813,"description":"Reka Flash 3 is a general-purpose, instruction-tuned large language model with 21 billion parameters, developed by Reka. It excels at general chat, coding tasks, instruction-following, and function calling. Featuring a...","context_length":65536,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000001","completion":"0.0000002"},"top_provider":{"context_length":65536,"max_completion_tokens":58982,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logprobs","max_tokens","presence_penalty","reasoning","seed","stop","structured_outputs","temperature","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2025-01-31","expiration_date":null,"links":{"details":"/api/v1/models/rekaai/reka-flash-3/endpoints"},"reasoning":{"mandatory":true}},{"id":"google/gemma-3-27b-it","canonical_slug":"google/gemma-3-27b-it","hugging_face_id":"google/gemma-3-27b-it","name":"Google: Gemma 3 27B","created":1741756359,"description":"Gemma 3 introduces multimodality, supporting vision-language input and text outputs. It handles context windows up to 128k tokens, understands over 140 languages, and offers improved math, reasoning, and chat capabilities,...","context_length":131072,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":"gemma"},"pricing":{"prompt":"0.00000008","completion":"0.00000045","input_cache_read":"0.00000004"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-08-31","expiration_date":null,"links":{"details":"/api/v1/models/google/gemma-3-27b-it/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":4.9,"coding_index":10.1,"agentic_index":0.1}}},{"id":"thedrummer/skyfall-36b-v2","canonical_slug":"thedrummer/skyfall-36b-v2","hugging_face_id":"TheDrummer/Skyfall-36B-v2","name":"TheDrummer: Skyfall 36B V2","created":1741636566,"description":"Skyfall 36B v2 is an enhanced iteration of Mistral Small 2501, specifically fine-tuned for improved creativity, nuanced writing, role-playing, and coherent storytelling.","context_length":32768,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000055","completion":"0.0000008","input_cache_read":"0.00000025"},"top_provider":{"context_length":32768,"max_completion_tokens":29491,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/thedrummer/skyfall-36b-v2/endpoints"}},{"id":"perplexity/sonar-reasoning-pro","canonical_slug":"perplexity/sonar-reasoning-pro","hugging_face_id":"","name":"Perplexity: Sonar Reasoning Pro","created":1741313308,"description":"Note: Sonar Pro pricing includes Perplexity search pricing. See [details here](https://docs.perplexity.ai/guides/pricing#detailed-pricing-breakdown-for-sonar-reasoning-pro-and-sonar-pro) Sonar Reasoning Pro is a premier reasoning model powered by DeepSeek R1 with Chain of Thought (CoT). Designed for...","context_length":128000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":"deepseek-r1"},"pricing":{"prompt":"0.000002","completion":"0.000008","web_search":"0.005"},"top_provider":{"context_length":128000,"max_completion_tokens":115200,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","temperature","top_k","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/perplexity/sonar-reasoning-pro/endpoints"},"reasoning":{"mandatory":false}},{"id":"perplexity/sonar-pro","canonical_slug":"perplexity/sonar-pro","hugging_face_id":"","name":"Perplexity: Sonar Pro","created":1741312423,"description":"Note: Sonar Pro pricing includes Perplexity search pricing. See [details here](https://docs.perplexity.ai/guides/pricing#detailed-pricing-breakdown-for-sonar-reasoning-pro-and-sonar-pro) For enterprises seeking more advanced capabilities, the Sonar Pro API can handle in-depth, multi-step queries with added extensibility, like...","context_length":200000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000003","completion":"0.000015","web_search":"0.005"},"top_provider":{"context_length":200000,"max_completion_tokens":8000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","temperature","top_k","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/perplexity/sonar-pro/endpoints"}},{"id":"perplexity/sonar-deep-research","canonical_slug":"perplexity/sonar-deep-research","hugging_face_id":"","name":"Perplexity: Sonar Deep Research","created":1741311246,"description":"Sonar Deep Research is a research-focused model designed for multi-step retrieval, synthesis, and reasoning across complex topics. It autonomously searches, reads, and evaluates sources, refining its approach as it gathers...","context_length":128000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":"deepseek-r1"},"pricing":{"prompt":"0.000002","completion":"0.000008","web_search":"0.005","internal_reasoning":"0.000003"},"top_provider":{"context_length":128000,"max_completion_tokens":115200,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","temperature","top_k","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/perplexity/sonar-deep-research/endpoints"},"reasoning":{"mandatory":false}},{"id":"mistralai/mistral-saba","canonical_slug":"mistralai/mistral-saba-2502","hugging_face_id":"","name":"Mistral: Saba","created":1739803239,"description":"Mistral Saba is a 24B-parameter language model specifically designed for the Middle East and South Asia, delivering accurate and contextually relevant responses while maintaining efficient performance. Trained on curated regional...","context_length":32768,"architecture":{"modality":"text+file->text","input_modalities":["text","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.0000006","input_cache_read":"0.00000002"},"top_provider":{"context_length":32768,"max_completion_tokens":26214,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2024-09-30","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-saba-2502/endpoints"}},{"id":"openai/o3-mini-high","canonical_slug":"openai/o3-mini-high-2025-01-31","hugging_face_id":"","name":"OpenAI: o3 Mini High","created":1739372611,"description":"OpenAI o3-mini-high is the same model as [o3-mini](/openai/o3-mini) with reasoning_effort set to high. o3-mini is a cost-efficient language model optimized for STEM reasoning tasks, particularly excelling in science, mathematics, and...","context_length":200000,"architecture":{"modality":"text+file->text","input_modalities":["text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000011","completion":"0.0000044","web_search":"0.01","input_cache_read":"0.00000055"},"top_provider":{"context_length":200000,"max_completion_tokens":100000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","reasoning_effort","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/o3-mini-high-2025-01-31/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":11,"coding_index":16.3,"agentic_index":0.9}},"reasoning":{"mandatory":true,"supported_efforts":["high"],"default_effort":"high"}},{"id":"aion-labs/aion-rp-llama-3.1-8b","canonical_slug":"aion-labs/aion-rp-llama-3.1-8b","hugging_face_id":"","name":"AionLabs: Aion-RP 1.0 (8B)","created":1738696718,"description":"Aion-RP-Llama-3.1-8B ranks the highest in the character evaluation portion of the RPBench-Auto benchmark, a roleplaying-specific variant of Arena-Hard-Auto, where LLMs evaluate each other’s responses. It is a fine-tuned base model...","context_length":32768,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000008","completion":"0.0000016"},"top_provider":{"context_length":32768,"max_completion_tokens":29491,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_tokens","temperature","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/aion-labs/aion-rp-llama-3.1-8b/endpoints"}},{"id":"qwen/qwen2.5-vl-72b-instruct","canonical_slug":"qwen/qwen2.5-vl-72b-instruct","hugging_face_id":"Qwen/Qwen2.5-VL-72B-Instruct","name":"Qwen: Qwen2.5 VL 72B Instruct","created":1738410311,"description":"Qwen2.5-VL is proficient in recognizing common objects such as flowers, birds, fish, and insects. It is also highly capable of analyzing texts, charts, icons, graphics, and layouts within images.","context_length":128000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.0000008","completion":"0.000001","input_cache_read":"0.0000004"},"top_provider":{"context_length":128000,"max_completion_tokens":115200,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen2.5-vl-72b-instruct/endpoints"}},{"id":"qwen/qwen-plus","canonical_slug":"qwen/qwen-plus-2025-01-25","hugging_face_id":"","name":"Qwen: Qwen-Plus","created":1738409840,"description":"Qwen-Plus, based on the Qwen2.5 foundation model, is a 131K context model with a balanced performance, speed, and cost combination.","context_length":1000000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":null},"pricing":{"prompt":"0.00000026","completion":"0.00000078","input_cache_read":"0.000000052","input_cache_write":"0.000000325","overrides":[{"min_prompt_tokens":256000,"prompt":"0.00000078","completion":"0.00000234","input_cache_read":"0.000000156","input_cache_write":"0.000000975"}]},"top_provider":{"context_length":1000000,"max_completion_tokens":32768,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2025-03-31","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen-plus-2025-01-25/endpoints"}},{"id":"openai/o3-mini","canonical_slug":"openai/o3-mini-2025-01-31","hugging_face_id":"","name":"OpenAI: o3 Mini","created":1738351721,"description":"OpenAI o3-mini is a cost-efficient language model optimized for STEM reasoning tasks, particularly excelling in science, mathematics, and coding. This model supports the `reasoning_effort` parameter, which can be set to...","context_length":200000,"architecture":{"modality":"text+file->text","input_modalities":["text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000011","completion":"0.0000044","web_search":"0.01","input_cache_read":"0.00000055"},"top_provider":{"context_length":200000,"max_completion_tokens":100000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/o3-mini-2025-01-31/endpoints"},"reasoning":{"mandatory":false}},{"id":"openai/o3-mini:batch","canonical_slug":"openai/o3-mini-2025-01-31","hugging_face_id":"","name":"OpenAI: o3 Mini (batch)","created":1738351721,"description":"OpenAI o3-mini is a cost-efficient language model optimized for STEM reasoning tasks, particularly excelling in science, mathematics, and coding. This model supports the `reasoning_effort` parameter, which can be set to...","context_length":200000,"architecture":{"modality":"text+file->text","input_modalities":["text","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000055","completion":"0.0000022","web_search":"0.01","input_cache_read":"0.000000275"},"top_provider":{"context_length":200000,"max_completion_tokens":100000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/o3-mini-2025-01-31/endpoints"},"reasoning":{"mandatory":false}},{"id":"mistralai/mistral-small-24b-instruct-2501","canonical_slug":"mistralai/mistral-small-24b-instruct-2501","hugging_face_id":"mistralai/Mistral-Small-24B-Instruct-2501","name":"Mistral: Mistral Small 3","created":1738255409,"description":"Mistral Small 3 is a 24B-parameter language model optimized for low-latency performance across common AI tasks. Released under the Apache 2.0 license, it features both pre-trained and instruction-tuned versions designed...","context_length":32768,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.00000005","completion":"0.00000008"},"top_provider":{"context_length":32768,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_p"],"default_parameters":{"temperature":0.3,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-small-24b-instruct-2501/endpoints"}},{"id":"perplexity/sonar","canonical_slug":"perplexity/sonar","hugging_face_id":"","name":"Perplexity: Sonar","created":1738013808,"description":"Sonar is lightweight, affordable, fast, and simple to use — now featuring citations and the ability to customize sources. It is designed for companies seeking to integrate lightweight question-and-answer features...","context_length":127072,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000001","web_search":"0.005"},"top_provider":{"context_length":127072,"max_completion_tokens":114364,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","temperature","top_k","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/perplexity/sonar/endpoints"}},{"id":"deepseek/deepseek-r1-distill-llama-70b","canonical_slug":"deepseek/deepseek-r1-distill-llama-70b","hugging_face_id":"deepseek-ai/DeepSeek-R1-Distill-Llama-70B","name":"DeepSeek: R1 Distill Llama 70B","created":1737663169,"description":"DeepSeek R1 Distill Llama 70B is a distilled large language model based on [Llama-3.3-70B-Instruct](/meta-llama/llama-3.3-70b-instruct), using outputs from [DeepSeek R1](/deepseek/deepseek-r1). The model combines advanced distillation techniques to achieve high performance across...","context_length":8192,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama3","instruct_type":"deepseek-r1"},"pricing":{"prompt":"0.0000008","completion":"0.0000008"},"top_provider":{"context_length":8192,"max_completion_tokens":7372,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","repetition_penalty","seed","stop","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-07-31","expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-r1-distill-llama-70b/endpoints"},"reasoning":{"mandatory":false}},{"id":"deepseek/deepseek-r1","canonical_slug":"deepseek/deepseek-r1","hugging_face_id":"deepseek-ai/DeepSeek-R1","name":"DeepSeek: R1","created":1737381095,"description":"DeepSeek R1 is here: Performance on par with [OpenAI o1](/openai/o1), but open-sourced and with fully open reasoning tokens. It's 671B parameters in size, with 37B active in an inference pass....","context_length":64000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":"deepseek-r1"},"pricing":{"prompt":"0.0000007","completion":"0.0000025"},"top_provider":{"context_length":64000,"max_completion_tokens":16000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","max_tokens","presence_penalty","reasoning","repetition_penalty","response_format","seed","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-07-31","expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-r1/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":11.4,"coding_index":24.6,"agentic_index":1.1}},"reasoning":{"mandatory":true}},{"id":"minimax/minimax-01","canonical_slug":"minimax/minimax-01","hugging_face_id":"MiniMaxAI/MiniMax-Text-01","name":"MiniMax: MiniMax-01","created":1736915462,"description":"MiniMax-01 is a combines MiniMax-Text-01 for text generation and MiniMax-VL-01 for image understanding. It has 456 billion parameters, with 45.9 billion parameters activated per inference, and can handle a context...","context_length":1000192,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.0000002","completion":"0.0000011"},"top_provider":{"context_length":1000192,"max_completion_tokens":900172,"is_moderated":false},"per_request_limits":null,"supported_parameters":["max_tokens","temperature","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-03-31","expiration_date":null,"links":{"details":"/api/v1/models/minimax/minimax-01/endpoints"}},{"id":"microsoft/phi-4","canonical_slug":"microsoft/phi-4","hugging_face_id":"microsoft/phi-4","name":"Microsoft: Phi 4","created":1736489872,"description":"[Microsoft Research](/microsoft) Phi-4 is designed to perform well in complex reasoning tasks and can operate efficiently in situations with limited memory or where quick responses are needed. At 14 billion...","context_length":16384,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Other","instruct_type":null},"pricing":{"prompt":"0.00000007","completion":"0.00000014"},"top_provider":{"context_length":16384,"max_completion_tokens":14745,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/microsoft/phi-4/endpoints"}},{"id":"deepseek/deepseek-chat","canonical_slug":"deepseek/deepseek-chat-v3","hugging_face_id":"deepseek-ai/DeepSeek-V3","name":"DeepSeek: DeepSeek V3","created":1735241320,"description":"DeepSeek-V3 is the latest model from the DeepSeek team, building upon the instruction following and coding abilities of the previous versions. Pre-trained on nearly 15 trillion tokens, the reported evaluations...","context_length":163840,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"DeepSeek","instruct_type":null},"pricing":{"prompt":"0.0000002574","completion":"0.0000010287"},"top_provider":{"context_length":128000,"max_completion_tokens":16000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-07-31","expiration_date":null,"links":{"details":"/api/v1/models/deepseek/deepseek-chat-v3/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":1115,"win_rate":50.5,"rank":79},{"arena":"models","category":"codecategories","elo":1123,"win_rate":48.4,"rank":89},{"arena":"models","category":"dataviz","elo":1104,"win_rate":50,"rank":92},{"arena":"models","category":"gamedev","elo":1076,"win_rate":43.7,"rank":94},{"arena":"models","category":"svg","elo":1001,"win_rate":38.8,"rank":83},{"arena":"models","category":"uicomponent","elo":1108,"win_rate":52.7,"rank":84},{"arena":"models","category":"website","elo":1132,"win_rate":48.5,"rank":88}]}},{"id":"sao10k/l3.3-euryale-70b","canonical_slug":"sao10k/l3.3-euryale-70b-v2.3","hugging_face_id":"Sao10K/L3.3-70B-Euryale-v2.3","name":"Sao10K: Llama 3.3 Euryale 70B","created":1734535928,"description":"Euryale L3.3 70B is a model focused on creative roleplay from [Sao10k](https://ko-fi.com/sao10k). It is the successor of [Euryale L3 70B v2.2](/models/sao10k/l3-euryale-70b).","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama3","instruct_type":"llama3"},"pricing":{"prompt":"0.00000065","completion":"0.00000075"},"top_provider":{"context_length":131072,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logprobs","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/sao10k/l3.3-euryale-70b-v2.3/endpoints"}},{"id":"openai/o1","canonical_slug":"openai/o1-2024-12-17","hugging_face_id":"","name":"OpenAI: o1","created":1734459999,"description":"The latest and strongest model family from OpenAI, o1 is designed to spend more time thinking before responding. The o1 model series is trained with large-scale reinforcement learning to reason...","context_length":200000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000015","completion":"0.00006","web_search":"0.01","input_cache_read":"0.0000075"},"top_provider":{"context_length":200000,"max_completion_tokens":100000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["include_reasoning","max_tokens","reasoning","response_format","seed","structured_outputs","tool_choice","tools"],"default_parameters":{"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/o1-2024-12-17/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":39.7,"agentic_index":null}},"reasoning":{"mandatory":false}},{"id":"cohere/command-r7b-12-2024","canonical_slug":"cohere/command-r7b-12-2024","hugging_face_id":"","name":"Cohere: Command R7B (12-2024)","created":1734158152,"description":"Command R7B (12-2024) is a small, fast update of the Command R+ model, delivered in December 2024. It excels at RAG, tool use, agents, and similar tasks requiring complex reasoning...","context_length":128000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Cohere","instruct_type":null},"pricing":{"prompt":"0.0000000375","completion":"0.00000015"},"top_provider":{"context_length":128000,"max_completion_tokens":4000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-08-31","expiration_date":null,"links":{"details":"/api/v1/models/cohere/command-r7b-12-2024/endpoints"}},{"id":"meta-llama/llama-3.3-70b-instruct","canonical_slug":"meta-llama/llama-3.3-70b-instruct","hugging_face_id":"meta-llama/Llama-3.3-70B-Instruct","name":"Meta: Llama 3.3 70B Instruct","created":1733506137,"description":"The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out). The Llama 3.3 instruction tuned text only model...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama3","instruct_type":"llama3"},"pricing":{"prompt":"0.0000001","completion":"0.00000032"},"top_provider":{"context_length":131072,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/meta-llama/llama-3.3-70b-instruct/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":11.9,"agentic_index":null}}},{"id":"amazon/nova-lite-v1","canonical_slug":"amazon/nova-lite-v1","hugging_face_id":"","name":"Amazon: Nova Lite 1.0","created":1733437363,"description":"Amazon Nova Lite 1.0 is a very low-cost multimodal model from Amazon that focused on fast processing of image, video, and text inputs to generate text output. Amazon Nova Lite...","context_length":300000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Nova","instruct_type":null},"pricing":{"prompt":"0.00000006","completion":"0.00000024"},"top_provider":{"context_length":300000,"max_completion_tokens":5120,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_tokens","stop","temperature","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-10-31","expiration_date":null,"links":{"details":"/api/v1/models/amazon/nova-lite-v1/endpoints"}},{"id":"amazon/nova-micro-v1","canonical_slug":"amazon/nova-micro-v1","hugging_face_id":"","name":"Amazon: Nova Micro 1.0","created":1733437237,"description":"Amazon Nova Micro 1.0 is a text-only model that delivers the lowest latency responses in the Amazon Nova family of models at a very low cost. With a context length...","context_length":128000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Nova","instruct_type":null},"pricing":{"prompt":"0.000000035","completion":"0.00000014"},"top_provider":{"context_length":128000,"max_completion_tokens":5120,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_tokens","stop","temperature","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-10-31","expiration_date":null,"links":{"details":"/api/v1/models/amazon/nova-micro-v1/endpoints"}},{"id":"amazon/nova-pro-v1","canonical_slug":"amazon/nova-pro-v1","hugging_face_id":"","name":"Amazon: Nova Pro 1.0","created":1733436303,"description":"Amazon Nova Pro 1.0 is a capable multimodal model from Amazon focused on providing a combination of accuracy, speed, and cost for a wide range of tasks. As of December...","context_length":300000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Nova","instruct_type":null},"pricing":{"prompt":"0.0000008","completion":"0.0000032"},"top_provider":{"context_length":300000,"max_completion_tokens":5120,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_tokens","stop","temperature","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-10-31","expiration_date":null,"links":{"details":"/api/v1/models/amazon/nova-pro-v1/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"website","elo":808,"win_rate":21.4,"rank":130}]}},{"id":"openai/gpt-4o-2024-11-20","canonical_slug":"openai/gpt-4o-2024-11-20","hugging_face_id":"","name":"OpenAI: GPT-4o (2024-11-20)","created":1732127594,"description":"The 2024-11-20 version of GPT-4o offers a leveled-up creative writing ability with more natural, engaging, and tailored writing to improve relevance & readability. It’s also better at working with uploaded...","context_length":128000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.00001","input_cache_read":"0.00000125"},"top_provider":{"context_length":128000,"max_completion_tokens":16384,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","prediction","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4o-2024-11-20/endpoints"}},{"id":"mistralai/mistral-large-2407","canonical_slug":"mistralai/mistral-large-2407","hugging_face_id":"","name":"Mistral Large 2407","created":1731978415,"description":"This is Mistral AI's flagship model, Mistral Large 2 (version mistral-large-2407). It's a proprietary weights-available model and excels at reasoning, code, JSON, chat, and more. Read the launch announcement [here](https://mistral.ai/news/mistral-large-2407/)....","context_length":131072,"architecture":{"modality":"text+file->text","input_modalities":["text","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000006","input_cache_read":"0.0000002"},"top_provider":{"context_length":131072,"max_completion_tokens":104857,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2024-03-31","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-large-2407/endpoints"}},{"id":"qwen/qwen-2.5-coder-32b-instruct","canonical_slug":"qwen/qwen-2.5-coder-32b-instruct","hugging_face_id":"Qwen/Qwen2.5-Coder-32B-Instruct","name":"Qwen2.5 Coder 32B Instruct","created":1731368400,"description":"Qwen2.5-Coder is the latest series of Code-Specific Qwen large language models (formerly known as CodeQwen). Qwen2.5-Coder brings the following improvements upon CodeQwen1.5: - Significantly improvements in **code generation**, **code reasoning**...","context_length":32768,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":"chatml"},"pricing":{"prompt":"0.00000066","completion":"0.000001"},"top_provider":{"context_length":32768,"max_completion_tokens":29491,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","seed","stop","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen-2.5-coder-32b-instruct/endpoints"}},{"id":"thedrummer/unslopnemo-12b","canonical_slug":"thedrummer/unslopnemo-12b","hugging_face_id":"TheDrummer/UnslopNemo-12B-v4.1","name":"TheDrummer: UnslopNemo 12B","created":1731103448,"description":"UnslopNemo v4.1 is the latest addition from the creator of Rocinante, designed for adventure writing and role-play scenarios.","context_length":1024000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":"mistral"},"pricing":{"prompt":"0.0000004","completion":"0.0000004"},"top_provider":{"context_length":1024000,"max_completion_tokens":819200,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-04-30","expiration_date":null,"links":{"details":"/api/v1/models/thedrummer/unslopnemo-12b/endpoints"}},{"id":"anthracite-org/magnum-v4-72b","canonical_slug":"anthracite-org/magnum-v4-72b","hugging_face_id":"anthracite-org/magnum-v4-72b","name":"Magnum v4 72B","created":1729555200,"description":"This is a series of models designed to replicate the prose quality of the Claude 3 models, specifically Sonnet(https://openrouter.ai/anthropic/claude-3.5-sonnet) and Opus(https://openrouter.ai/anthropic/claude-3-opus).\n\nThe model is fine-tuned on top of [Qwen2.5 72B](https://openrouter.ai/qwen/qwen-2.5-72b-instruct).","context_length":32768,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":"chatml"},"pricing":{"prompt":"0.0000025","completion":"0.000005"},"top_provider":{"context_length":32768,"max_completion_tokens":4096,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_a","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/anthracite-org/magnum-v4-72b/endpoints"}},{"id":"qwen/qwen-2.5-7b-instruct","canonical_slug":"qwen/qwen-2.5-7b-instruct","hugging_face_id":"Qwen/Qwen2.5-7B-Instruct","name":"Qwen: Qwen2.5 7B Instruct","created":1729036800,"description":"Qwen2.5 7B is the latest series of Qwen large language models. Qwen2.5 brings the following improvements upon Qwen2: - Significantly more knowledge and has greatly improved capabilities in coding and...","context_length":32768,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":"chatml"},"pricing":{"prompt":"0.0000001","completion":"0.0000002"},"top_provider":{"context_length":32768,"max_completion_tokens":29491,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{"temperature":null,"top_p":null,"frequency_penalty":null},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen-2.5-7b-instruct/endpoints"}},{"id":"meta-llama/llama-3.2-1b-instruct","canonical_slug":"meta-llama/llama-3.2-1b-instruct","hugging_face_id":"meta-llama/Llama-3.2-1B-Instruct","name":"Meta: Llama 3.2 1B Instruct","created":1727222400,"description":"Llama 3.2 1B is a 1-billion-parameter language model focused on efficiently performing natural language tasks, such as summarization, dialogue, and multilingual text analysis. Its smaller size allows it to operate...","context_length":60000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama3","instruct_type":"llama3"},"pricing":{"prompt":"0.000000027","completion":"0.000000201"},"top_provider":{"context_length":60000,"max_completion_tokens":54000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","seed","stop","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/meta-llama/llama-3.2-1b-instruct/endpoints"}},{"id":"meta-llama/llama-3.2-3b-instruct","canonical_slug":"meta-llama/llama-3.2-3b-instruct","hugging_face_id":"meta-llama/Llama-3.2-3B-Instruct","name":"Meta: Llama 3.2 3B Instruct","created":1727222400,"description":"Llama 3.2 3B is a 3-billion-parameter multilingual large language model, optimized for advanced natural language processing tasks like dialogue generation, reasoning, and summarization. Designed with the latest transformer architecture, it...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama3","instruct_type":"llama3"},"pricing":{"prompt":"0.00000005","completion":"0.00000033"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","seed","stop","structured_outputs","temperature","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/meta-llama/llama-3.2-3b-instruct/endpoints"}},{"id":"qwen/qwen-2.5-72b-instruct","canonical_slug":"qwen/qwen-2.5-72b-instruct","hugging_face_id":"Qwen/Qwen2.5-72B-Instruct","name":"Qwen2.5 72B Instruct","created":1726704000,"description":"Qwen2.5 72B is the latest series of Qwen large language models. Qwen2.5 brings the following improvements upon Qwen2: - Significantly more knowledge and has greatly improved capabilities in coding and...","context_length":32768,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Qwen","instruct_type":"chatml"},"pricing":{"prompt":"0.00000036","completion":"0.0000004"},"top_provider":{"context_length":32768,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/qwen/qwen-2.5-72b-instruct/endpoints"}},{"id":"cohere/command-r-08-2024","canonical_slug":"cohere/command-r-08-2024","hugging_face_id":null,"name":"Cohere: Command R (08-2024)","created":1724976000,"description":"command-r-08-2024 is an update of the [Command R](/models/cohere/command-r) with improved performance for multilingual retrieval-augmented generation (RAG) and tool use. More broadly, it is better at math, code and reasoning and...","context_length":128000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Cohere","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000006"},"top_provider":{"context_length":128000,"max_completion_tokens":4000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-03-31","expiration_date":null,"links":{"details":"/api/v1/models/cohere/command-r-08-2024/endpoints"}},{"id":"cohere/command-r-plus-08-2024","canonical_slug":"cohere/command-r-plus-08-2024","hugging_face_id":null,"name":"Cohere: Command R+ (08-2024)","created":1724976000,"description":"command-r-plus-08-2024 is an update of the [Command R+](/models/cohere/command-r-plus) with roughly 50% higher throughput and 25% lower latencies as compared to the previous Command R+ version, while keeping the hardware footprint...","context_length":128000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Cohere","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.00001"},"top_provider":{"context_length":128000,"max_completion_tokens":4000,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-03-31","expiration_date":null,"links":{"details":"/api/v1/models/cohere/command-r-plus-08-2024/endpoints"}},{"id":"sao10k/l3.1-euryale-70b","canonical_slug":"sao10k/l3.1-euryale-70b","hugging_face_id":"Sao10K/L3.1-70B-Euryale-v2.2","name":"Sao10K: Llama 3.1 Euryale 70B v2.2","created":1724803200,"description":"Euryale L3.1 70B v2.2 is a model focused on creative roleplay from [Sao10k](https://ko-fi.com/sao10k). It is the successor of [Euryale L3 70B v2.1](/models/sao10k/l3-euryale-70b).","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama3","instruct_type":"llama3"},"pricing":{"prompt":"0.00000085","completion":"0.00000085"},"top_provider":{"context_length":131072,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/sao10k/l3.1-euryale-70b/endpoints"}},{"id":"nousresearch/hermes-3-llama-3.1-70b","canonical_slug":"nousresearch/hermes-3-llama-3.1-70b","hugging_face_id":"NousResearch/Hermes-3-Llama-3.1-70B","name":"Nous: Hermes 3 70B Instruct","created":1723939200,"description":"Hermes 3 is a generalist language model with many improvements over [Hermes 2](/models/nousresearch/nous-hermes-2-mistral-7b-dpo), including advanced agentic capabilities, much better roleplaying, reasoning, multi-turn conversation, long context coherence, and improvements across the...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama3","instruct_type":"chatml"},"pricing":{"prompt":"0.0000007","completion":"0.0000007"},"top_provider":{"context_length":131072,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/nousresearch/hermes-3-llama-3.1-70b/endpoints"}},{"id":"nousresearch/hermes-3-llama-3.1-405b","canonical_slug":"nousresearch/hermes-3-llama-3.1-405b","hugging_face_id":"NousResearch/Hermes-3-Llama-3.1-405B","name":"Nous: Hermes 3 405B Instruct","created":1723766400,"description":"Hermes 3 is a generalist language model with many improvements over Hermes 2, including advanced agentic capabilities, much better roleplaying, reasoning, multi-turn conversation, long context coherence, and improvements across the...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama3","instruct_type":"chatml"},"pricing":{"prompt":"0.000001","completion":"0.000001"},"top_provider":{"context_length":131072,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/nousresearch/hermes-3-llama-3.1-405b/endpoints"}},{"id":"sao10k/l3-lunaris-8b","canonical_slug":"sao10k/l3-lunaris-8b","hugging_face_id":"Sao10K/L3-8B-Lunaris-v1","name":"Sao10K: Llama 3 8B Lunaris","created":1723507200,"description":"Lunaris 8B is a versatile generalist and roleplaying model based on Llama 3. It's a strategic merge of multiple models, designed to balance creativity with improved logic and general knowledge....","context_length":8192,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama3","instruct_type":"llama3"},"pricing":{"prompt":"0.00000004","completion":"0.00000005"},"top_provider":{"context_length":8192,"max_completion_tokens":7372,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/sao10k/l3-lunaris-8b/endpoints"}},{"id":"openai/gpt-4o-2024-08-06","canonical_slug":"openai/gpt-4o-2024-08-06","hugging_face_id":null,"name":"OpenAI: GPT-4o (2024-08-06)","created":1722902400,"description":"The 2024-08-06 version of GPT-4o offers improved performance in structured outputs, with the ability to supply a JSON schema in the respone_format. Read more [here](https://openai.com/index/introducing-structured-outputs-in-the-api/). GPT-4o (\"o\" for \"omni\") is...","context_length":128000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.00001","input_cache_read":"0.00000125"},"top_provider":{"context_length":128000,"max_completion_tokens":16384,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_completion_tokens","max_tokens","prediction","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4o-2024-08-06/endpoints"}},{"id":"meta-llama/llama-3.1-70b-instruct","canonical_slug":"meta-llama/llama-3.1-70b-instruct","hugging_face_id":"meta-llama/Meta-Llama-3.1-70B-Instruct","name":"Meta: Llama 3.1 70B Instruct","created":1721692800,"description":"Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors. This 70B instruct-tuned version is optimized for high quality dialogue usecases. It has demonstrated strong...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama3","instruct_type":"llama3"},"pricing":{"prompt":"0.0000004","completion":"0.0000004"},"top_provider":{"context_length":131072,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/meta-llama/llama-3.1-70b-instruct/endpoints"}},{"id":"meta-llama/llama-3.1-8b-instruct","canonical_slug":"meta-llama/llama-3.1-8b-instruct","hugging_face_id":"meta-llama/Meta-Llama-3.1-8B-Instruct","name":"Meta: Llama 3.1 8B Instruct","created":1721692800,"description":"Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors. This 8B instruct-tuned version is fast and efficient. It has demonstrated strong performance compared to...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama3","instruct_type":"llama3"},"pricing":{"prompt":"0.00000005","completion":"0.00000008","input_cache_read":"0.000000025"},"top_provider":{"context_length":131072,"max_completion_tokens":117964,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/meta-llama/llama-3.1-8b-instruct/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":5.4,"agentic_index":null}}},{"id":"mistralai/mistral-nemo","canonical_slug":"mistralai/mistral-nemo","hugging_face_id":"mistralai/Mistral-Nemo-Instruct-2407","name":"Mistral: Mistral Nemo","created":1721347200,"description":"A 12B parameter model with a 128k token context length built by Mistral in collaboration with NVIDIA. The model is multilingual, supporting English, French, German, Spanish, Italian, Portuguese, Chinese, Japanese,...","context_length":131072,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.000000019","completion":"0.00000003"},"top_provider":{"context_length":131072,"max_completion_tokens":16384,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_k","top_logprobs","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2024-04-30","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-nemo/endpoints"}},{"id":"openai/gpt-4o-mini","canonical_slug":"openai/gpt-4o-mini","hugging_face_id":null,"name":"OpenAI: GPT-4o-mini","created":1721260800,"description":"GPT-4o mini is OpenAI's newest model after [GPT-4 Omni](/models/openai/gpt-4o), supporting both text and image inputs with text outputs. As their most advanced small model, it is many multiples more affordable...","context_length":128000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000006","input_cache_read":"0.000000075"},"top_provider":{"context_length":128000,"max_completion_tokens":16384,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_completion_tokens","max_tokens","prediction","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4o-mini/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":11.4,"agentic_index":null}}},{"id":"openai/gpt-4o-mini-2024-07-18","canonical_slug":"openai/gpt-4o-mini-2024-07-18","hugging_face_id":null,"name":"OpenAI: GPT-4o-mini (2024-07-18)","created":1721260800,"description":"GPT-4o mini is OpenAI's newest model after [GPT-4 Omni](/models/openai/gpt-4o), supporting both text and image inputs with text outputs. As their most advanced small model, it is many multiples more affordable...","context_length":128000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000015","completion":"0.0000006","input_cache_read":"0.000000075"},"top_provider":{"context_length":128000,"max_completion_tokens":16384,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","prediction","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4o-mini-2024-07-18/endpoints"}},{"id":"openai/gpt-4o-mini:batch","canonical_slug":"openai/gpt-4o-mini","hugging_face_id":null,"name":"OpenAI: GPT-4o-mini (batch)","created":1721260800,"description":"GPT-4o mini is OpenAI's newest model after [GPT-4 Omni](/models/openai/gpt-4o), supporting both text and image inputs with text outputs. As their most advanced small model, it is many multiples more affordable...","context_length":128000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000000075","completion":"0.0000003","web_search":"0.01","input_cache_read":"0.0000000375"},"top_provider":{"context_length":128000,"max_completion_tokens":16384,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","prediction","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4o-mini/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":11.4,"agentic_index":null}}},{"id":"google/gemma-2-27b-it","canonical_slug":"google/gemma-2-27b-it","hugging_face_id":"google/gemma-2-27b-it","name":"Google: Gemma 2 27B","created":1720828800,"description":"Gemma 2 27B by Google is an open model built from the same research and technology used to create the [Gemini models](/models?q=gemini). Gemma models are well-suited for a variety of...","context_length":8192,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Gemini","instruct_type":"gemma"},"pricing":{"prompt":"0.00000065","completion":"0.00000065"},"top_provider":{"context_length":8192,"max_completion_tokens":2048,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-06-30","expiration_date":null,"links":{"details":"/api/v1/models/google/gemma-2-27b-it/endpoints"}},{"id":"openai/gpt-4o","canonical_slug":"openai/gpt-4o","hugging_face_id":null,"name":"OpenAI: GPT-4o","created":1715558400,"description":"GPT-4o (\"o\" for \"omni\") is OpenAI's latest AI model, supporting both text and image inputs with text outputs. It maintains the intelligence level of [GPT-4 Turbo](/models/openai/gpt-4-turbo) while being twice as...","context_length":128000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000025","completion":"0.00001","input_cache_read":"0.00000125"},"top_provider":{"context_length":128000,"max_completion_tokens":16384,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_completion_tokens","max_tokens","prediction","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4o/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":897,"win_rate":39.2,"rank":111},{"arena":"models","category":"codecategories","elo":875,"win_rate":34.8,"rank":123},{"arena":"models","category":"dataviz","elo":871,"win_rate":36,"rank":120},{"arena":"models","category":"gamedev","elo":929,"win_rate":42.3,"rank":119},{"arena":"models","category":"uicomponent","elo":900,"win_rate":38.1,"rank":117},{"arena":"models","category":"website","elo":843,"win_rate":31.5,"rank":129}]}},{"id":"openai/gpt-4o-2024-05-13","canonical_slug":"openai/gpt-4o-2024-05-13","hugging_face_id":null,"name":"OpenAI: GPT-4o (2024-05-13)","created":1715558400,"description":"GPT-4o (\"o\" for \"omni\") is OpenAI's latest AI model, supporting both text and image inputs with text outputs. It maintains the intelligence level of [GPT-4 Turbo](/models/openai/gpt-4-turbo) while being twice as...","context_length":128000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000015"},"top_provider":{"context_length":128000,"max_completion_tokens":4096,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_completion_tokens","max_tokens","prediction","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4o-2024-05-13/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":24.2,"agentic_index":null}}},{"id":"openai/gpt-4o:batch","canonical_slug":"openai/gpt-4o","hugging_face_id":null,"name":"OpenAI: GPT-4o (batch)","created":1715558400,"description":"GPT-4o (\"o\" for \"omni\") is OpenAI's latest AI model, supporting both text and image inputs with text outputs. It maintains the intelligence level of [GPT-4 Turbo](/models/openai/gpt-4-turbo) while being twice as...","context_length":128000,"architecture":{"modality":"text+image+file->text","input_modalities":["text","image","file"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000125","completion":"0.000005","web_search":"0.01","input_cache_read":"0.000000625"},"top_provider":{"context_length":128000,"max_completion_tokens":16384,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","prediction","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-10-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4o/endpoints"},"benchmarks":{"design_arena":[{"arena":"models","category":"3d","elo":897,"win_rate":39.2,"rank":111},{"arena":"models","category":"codecategories","elo":875,"win_rate":34.8,"rank":123},{"arena":"models","category":"dataviz","elo":871,"win_rate":36,"rank":120},{"arena":"models","category":"gamedev","elo":929,"win_rate":42.3,"rank":119},{"arena":"models","category":"uicomponent","elo":900,"win_rate":38.1,"rank":117},{"arena":"models","category":"website","elo":843,"win_rate":31.5,"rank":129}]}},{"id":"mistralai/mixtral-8x22b-instruct","canonical_slug":"mistralai/mixtral-8x22b-instruct","hugging_face_id":"mistralai/Mixtral-8x22B-Instruct-v0.1","name":"Mistral: Mixtral 8x22B Instruct","created":1713312000,"description":"Mistral's official instruct fine-tuned version of [Mixtral 8x22B](/models/mistralai/mixtral-8x22b). It uses 39B active parameters out of 141B, offering unparalleled cost efficiency for its size. Its strengths include: - strong math, coding,...","context_length":65536,"architecture":{"modality":"text+file->text","input_modalities":["text","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":"mistral"},"pricing":{"prompt":"0.000002","completion":"0.000006","input_cache_read":"0.0000002"},"top_provider":{"context_length":65536,"max_completion_tokens":52428,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2024-01-31","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mixtral-8x22b-instruct/endpoints"}},{"id":"microsoft/wizardlm-2-8x22b","canonical_slug":"microsoft/wizardlm-2-8x22b","hugging_face_id":"microsoft/WizardLM-2-8x22B","name":"WizardLM-2 8x22B","created":1713225600,"description":"WizardLM-2 8x22B is Microsoft AI's most advanced Wizard model. It demonstrates highly competitive performance compared to leading proprietary models, and it consistently outperforms all existing state-of-the-art opensource models. It is...","context_length":65535,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":"vicuna"},"pricing":{"prompt":"0.00000062","completion":"0.00000062"},"top_provider":{"context_length":65535,"max_completion_tokens":8000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","repetition_penalty","response_format","seed","stop","temperature","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2024-04-30","expiration_date":null,"links":{"details":"/api/v1/models/microsoft/wizardlm-2-8x22b/endpoints"}},{"id":"openai/gpt-4-turbo","canonical_slug":"openai/gpt-4-turbo","hugging_face_id":null,"name":"OpenAI: GPT-4 Turbo","created":1712620800,"description":"The latest GPT-4 Turbo model with vision capabilities. Vision requests can now use JSON mode and function calling.\n\nTraining data: up to December 2023.","context_length":128000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00001","completion":"0.00003"},"top_provider":{"context_length":128000,"max_completion_tokens":4096,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4-turbo/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":21.5,"agentic_index":null}}},{"id":"openai/gpt-4-turbo:batch","canonical_slug":"openai/gpt-4-turbo","hugging_face_id":null,"name":"OpenAI: GPT-4 Turbo (batch)","created":1712620800,"description":"The latest GPT-4 Turbo model with vision capabilities. Vision requests can now use JSON mode and function calling.\n\nTraining data: up to December 2023.","context_length":128000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000005","completion":"0.000015","web_search":"0.01"},"top_provider":{"context_length":128000,"max_completion_tokens":4096,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4-turbo/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":21.5,"agentic_index":null}}},{"id":"anthropic/claude-3-haiku","canonical_slug":"anthropic/claude-3-haiku","hugging_face_id":null,"name":"Anthropic: Claude 3 Haiku","created":1710288000,"description":"Claude 3 Haiku is Anthropic's fastest and most compact model for\nnear-instant responsiveness. Quick and accurate targeted performance.\n\nSee the launch announcement and benchmark results [here](https://www.anthropic.com/news/claude-3-haiku)\n\n#multimodal","context_length":200000,"architecture":{"modality":"text+image->text","input_modalities":["text","image"],"output_modalities":["text"],"tokenizer":"Claude","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.00000125","web_search":"0.01","input_cache_read":"0.00000003","input_cache_write":"0.0000003","input_cache_write_1h":"0.0000005"},"top_provider":{"context_length":200000,"max_completion_tokens":4096,"is_moderated":true},"per_request_limits":null,"supported_parameters":["max_tokens","stop","temperature","tool_choice","tools","top_k","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-08-31","expiration_date":null,"links":{"details":"/api/v1/models/anthropic/claude-3-haiku/endpoints"}},{"id":"mistralai/mistral-large","canonical_slug":"mistralai/mistral-large","hugging_face_id":null,"name":"Mistral Large","created":1708905600,"description":"This is Mistral AI's flagship model, Mistral Large 2 (version `mistral-large-2407`). It's a proprietary weights-available model and excels at reasoning, code, JSON, chat, and more. Read the launch announcement [here](https://mistral.ai/news/mistral-large-2407/)....","context_length":128000,"architecture":{"modality":"text+file->text","input_modalities":["text","file"],"output_modalities":["text"],"tokenizer":"Mistral","instruct_type":null},"pricing":{"prompt":"0.000002","completion":"0.000006","input_cache_read":"0.0000002"},"top_provider":{"context_length":128000,"max_completion_tokens":102400,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_p"],"default_parameters":{"temperature":0.3},"supported_voices":null,"knowledge_cutoff":"2024-11-30","expiration_date":null,"links":{"details":"/api/v1/models/mistralai/mistral-large/endpoints"}},{"id":"openai/gpt-3.5-turbo-0613","canonical_slug":"openai/gpt-3.5-turbo-0613","hugging_face_id":null,"name":"OpenAI: GPT-3.5 Turbo (older v0613)","created":1706140800,"description":"GPT-3.5 Turbo is OpenAI's fastest model. It can understand and generate natural language or code, and is optimized for chat and traditional completion tasks.\n\nTraining data up to Sep 2021.","context_length":4095,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000001","completion":"0.000002"},"top_provider":{"context_length":4095,"max_completion_tokens":3685,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_completion_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2021-09-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-3.5-turbo-0613/endpoints"}},{"id":"openai/gpt-4-turbo-preview","canonical_slug":"openai/gpt-4-turbo-preview","hugging_face_id":null,"name":"OpenAI: GPT-4 Turbo Preview","created":1706140800,"description":"The preview GPT-4 model with improved instruction following, JSON mode, reproducible outputs, parallel function calling, and more. Training data: up to Dec 2023. **Note:** heavily rate limited by OpenAI while...","context_length":128000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00001","completion":"0.00003"},"top_provider":{"context_length":128000,"max_completion_tokens":4096,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-12-31","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4-turbo-preview/endpoints"}},{"id":"openrouter/auto","canonical_slug":"openrouter/auto","hugging_face_id":null,"name":"Auto Router","created":1699401600,"description":"The Auto Router automatically selects the best model for your prompt, powered by the wisdom of the market. It routes you based on what the OpenRouter community collectively spends on...","context_length":2000000,"architecture":{"modality":"text+image+file+audio+video->text+image","input_modalities":["text","image","audio","file","video"],"output_modalities":["text","image"],"tokenizer":"Router","instruct_type":null},"pricing":{"prompt":"-1","completion":"-1"},"top_provider":{"context_length":null,"max_completion_tokens":null,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","include_reasoning","logit_bias","logprobs","max_tokens","min_p","prediction","presence_penalty","reasoning","reasoning_effort","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_a","top_k","top_logprobs","top_p","web_search_options"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":null,"expiration_date":null,"links":{"details":"/api/v1/models/openrouter/auto/endpoints"}},{"id":"openai/gpt-3.5-turbo-instruct","canonical_slug":"openai/gpt-3.5-turbo-instruct","hugging_face_id":null,"name":"OpenAI: GPT-3.5 Turbo Instruct","created":1695859200,"description":"This model is a variant of GPT-3.5 Turbo tuned for instructional prompts and omitting chat-related optimizations. Training data: up to Sep 2021.","context_length":4095,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":"chatml"},"pricing":{"prompt":"0.0000015","completion":"0.000002"},"top_provider":{"context_length":4095,"max_completion_tokens":3685,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2021-09-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-3.5-turbo-instruct/endpoints"}},{"id":"openai/gpt-3.5-turbo-16k","canonical_slug":"openai/gpt-3.5-turbo-16k","hugging_face_id":null,"name":"OpenAI: GPT-3.5 Turbo 16k","created":1693180800,"description":"This model offers four times the context length of gpt-3.5-turbo, allowing it to support approximately 20 pages of text in a single request at a higher cost. Training data: up...","context_length":16385,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.000003","completion":"0.000004"},"top_provider":{"context_length":16385,"max_completion_tokens":4096,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_completion_tokens","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2021-09-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-3.5-turbo-16k/endpoints"}},{"id":"mancer/weaver","canonical_slug":"mancer/weaver","hugging_face_id":null,"name":"Mancer: Weaver (alpha)","created":1690934400,"description":"An attempt to recreate Claude-style verbosity, but don't expect the same level of coherence or memory. Meant for use in roleplay/narrative situations.","context_length":8000,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama2","instruct_type":"alpaca"},"pricing":{"prompt":"0.0000004","completion":"0.00000075"},"top_provider":{"context_length":8000,"max_completion_tokens":6000,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","temperature","top_a","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-06-30","expiration_date":null,"links":{"details":"/api/v1/models/mancer/weaver/endpoints"}},{"id":"undi95/remm-slerp-l2-13b","canonical_slug":"undi95/remm-slerp-l2-13b","hugging_face_id":"Undi95/ReMM-SLERP-L2-13B","name":"ReMM SLERP 13B","created":1689984000,"description":"A recreation trial of the original MythoMax-L2-B13 but with updated models. #merge","context_length":6144,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama2","instruct_type":"alpaca"},"pricing":{"prompt":"0.00000035","completion":"0.00000065"},"top_provider":{"context_length":6144,"max_completion_tokens":5529,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_a","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-06-30","expiration_date":null,"links":{"details":"/api/v1/models/undi95/remm-slerp-l2-13b/endpoints"}},{"id":"gryphe/mythomax-l2-13b","canonical_slug":"gryphe/mythomax-l2-13b","hugging_face_id":"Gryphe/MythoMax-L2-13b","name":"MythoMax 13B","created":1688256000,"description":"One of the highest performing and most popular fine-tunes of Llama 2 13B, with rich descriptions and roleplay. #merge","context_length":8192,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"Llama2","instruct_type":"alpaca"},"pricing":{"prompt":"0.00000006","completion":"0.00000006"},"top_provider":{"context_length":4096,"max_completion_tokens":3686,"is_moderated":false},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","min_p","presence_penalty","repetition_penalty","response_format","seed","stop","structured_outputs","temperature","top_a","top_k","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2023-06-30","expiration_date":null,"links":{"details":"/api/v1/models/gryphe/mythomax-l2-13b/endpoints"}},{"id":"openai/gpt-3.5-turbo","canonical_slug":"openai/gpt-3.5-turbo","hugging_face_id":null,"name":"OpenAI: GPT-3.5 Turbo","created":1685232000,"description":"GPT-3.5 Turbo is OpenAI's fastest model. It can understand and generate natural language or code, and is optimized for chat and traditional completion tasks.\n\nTraining data up to Sep 2021.","context_length":16385,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.0000005","completion":"0.0000015"},"top_provider":{"context_length":16385,"max_completion_tokens":4096,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2021-09-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-3.5-turbo/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":10.7,"agentic_index":null}}},{"id":"openai/gpt-3.5-turbo:batch","canonical_slug":"openai/gpt-3.5-turbo","hugging_face_id":null,"name":"OpenAI: GPT-3.5 Turbo (batch)","created":1685232000,"description":"GPT-3.5 Turbo is OpenAI's fastest model. It can understand and generate natural language or code, and is optimized for chat and traditional completion tasks.\n\nTraining data up to Sep 2021.","context_length":16385,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00000025","completion":"0.00000075","web_search":"0.01"},"top_provider":{"context_length":16385,"max_completion_tokens":4096,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2021-09-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-3.5-turbo/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":10.7,"agentic_index":null}}},{"id":"openai/gpt-4","canonical_slug":"openai/gpt-4","hugging_face_id":null,"name":"OpenAI: GPT-4","created":1685232000,"description":"OpenAI's flagship model, GPT-4 is a large-scale multimodal language model capable of solving difficult problems with greater accuracy than previous models due to its broader general knowledge and advanced reasoning...","context_length":8191,"architecture":{"modality":"text->text","input_modalities":["text"],"output_modalities":["text"],"tokenizer":"GPT","instruct_type":null},"pricing":{"prompt":"0.00003","completion":"0.00006"},"top_provider":{"context_length":8191,"max_completion_tokens":4096,"is_moderated":true},"per_request_limits":null,"supported_parameters":["frequency_penalty","logit_bias","logprobs","max_completion_tokens","max_tokens","presence_penalty","response_format","seed","stop","structured_outputs","temperature","tool_choice","tools","top_logprobs","top_p"],"default_parameters":{},"supported_voices":null,"knowledge_cutoff":"2021-09-30","expiration_date":null,"links":{"details":"/api/v1/models/openai/gpt-4/endpoints"},"benchmarks":{"design_arena":[],"artificial_analysis":{"intelligence_index":null,"coding_index":13.1,"agentic_index":null}}}],"total_count":447,"links":{"next":null}} \ No newline at end of file diff --git a/data/artifacts/471e5e2ccda3459a0170200fbcac5b7f9a18f45ae7e4fa10f799e7eda7d625a5 b/data/artifacts/471e5e2ccda3459a0170200fbcac5b7f9a18f45ae7e4fa10f799e7eda7d625a5 new file mode 100644 index 0000000..4d363c2 --- /dev/null +++ b/data/artifacts/471e5e2ccda3459a0170200fbcac5b7f9a18f45ae7e4fa10f799e7eda7d625a5 @@ -0,0 +1,2 @@ +
Contact salesTry Claude

Free

Try Claude

$0

Free for everyone

Try Claude
  • Chat on web, iOS, Android, and on your desktop
  • Generate code and visualize data
  • Write, edit, and create content
  • Ability to search the web
  • Memory across conversations
  • Create files and execute code
  • Unlock more from Claude with desktop extensions
  • Connect Slack and Google Workspace services
  • Integrate any context or tool through connectors with remote MCP
  • Extended thinking for complex work

Pro

For everyday productivity

$17

Per month with annual subscription discount ($200 billed up front). $20 if billed monthly.

Try Claude

Everything in Free, plus:

  • More usage*
  • Includes Claude Code
  • Includes Claude Cowork
  • Includes Claude Design
  • Includes Claude Science
  • Access to unlimited projects to organize chats and documents
  • Access to Research
  • Ability to use more Claude models
  • Claude for Microsoft 365

Max

Get the most out of Claude

From $100

Per month

Try Claude

Everything in Pro, plus:

  • Choose 5x or 20x more usage than Pro*
  • Higher output limits for all tasks
  • Early access to advanced Claude features
  • Priority access at high traffic times

Usage limits apply. Prices shown don’t include applicable tax. Price and plans are subject to change at Anthropic's discretion.

Free
Try Claude
Pro
Try Claude
Max 5x
Try Claude
Max 20x
Try Claude

Features and capabilities

FeaturesFreeProMax 5xMax 20x

Chat on web, iOS, Android and Desktop

Yes
Yes
Yes
Yes

Claude Code

No
Yes
Yes
Yes

Claude Cowork

No
Yes
Yes
Yes

Claude Design

No
Yes
Yes
Yes

Claude Science

No
Yes
Yes
Yes

@Claude

No
No
No
No

Ability to search the web

Yes
Yes
Yes
Yes

Desktop extensions

Yes
Yes
Yes
Yes

Voice mode

Yes
Yes
Yes
Yes

Incognito chats

Yes
Yes
Yes
Yes

User preferences

Yes
Yes
Yes
Yes

Artifacts

Yes
Yes
Yes
Yes

Projects

No
Yes
Yes
Yes

Project sharing and collaboration

No
No
No
No

Create and edit files with code execution

Yes
Yes
Yes
Yes

Research

No
Yes
Yes
Yes

Memory

Yes
Yes
Yes
Yes

Skills

Yes
Yes
Yes
Yes

Connectors

Yes
Yes
Yes
Yes

Web search

Yes
Yes
Yes
Yes

Enterprise search

No
No
No
No

Claude in Chrome

No
Yes
Yes
Yes

Claude for Microsoft 365

No
Yes
Yes
Yes

Claude for Microsoft Outlook

No
Yes
Yes
Yes

Security and administration

FeaturesFreeProMax 5xMax 20x

Single sign-on (SSO) and domain capture

No
No
No
No

Central billing and administration

No
No
No
No

Admin controls for remote and local connectors

No
No
No
No

Enterprise deployment for the Claude desktop app

No
No
No
No

Role-based access

No
No
No
No

System for Cross-domain Identity Management (SCIM)

No
No
No
No

Audit logs

No
No
No
No

Usage analytics

No
No
No
No

Compliance API

No
No
No
No

HIPAA-ready offering

No
No
No
No

Custom data retention controls

No
No
No
No

Organization wide skills deployment

No
No
No
No

Payment options

FeaturesFreeProMax 5xMax 20x

Credit card

n/a
Yes
Yes
Yes

ACH

No
No
No
No

Invoicing / net terms

No
No
No
No

Billing cycle

n/aMonthly and annualMonthly and annualMonthly and annual

Adding seats midterm

n/an/an/an/a

Models and usage

FeaturesFreeProMax 5xMax 20x

Fable

No		Usage credits50% of weekly limits*50% of weekly limits*

Opus

No
Yes
Yes
Yes

Sonnet

Yes
Yes
Yes
Yes

Haiku

Yes
Yes
Yes
Yes

Context window

200k200k200k200k

Usage credits

No
Yes
Yes
Yes

Priority access at high traffic times

No
No
Yes
Yes

User and organizational level spend controls

No
No
No
No

Model training

Opt-outOpt-outOpt-outOpt-out

Plans and usage

Plans & Pricing | Claude by Anthropic \ No newline at end of file diff --git a/data/captures/34e27d084a98f664d939c0c58d5fc61876397d1436f711c01266647d1fe7e818.json b/data/captures/34e27d084a98f664d939c0c58d5fc61876397d1436f711c01266647d1fe7e818.json new file mode 100644 index 0000000..4cc85cf --- /dev/null +++ b/data/captures/34e27d084a98f664d939c0c58d5fc61876397d1436f711c01266647d1fe7e818.json @@ -0,0 +1 @@ +{"attempts":1,"bytes":264372,"changed":false,"content_sha256":"87cdcea7dd7bfc4e554d3fe4e8d3377ca8a47110130537414c82cc9ac61fdd80","final_url":"https://www.anthropic.com/news/claude-sonnet-5","media_type":"text/html; charset=utf-8","publisher":"Anthropic","retrieved_at":"2026-09-14T20:01:57.781138+00:00","schema_version":1,"source_id":"anthropic-sonnet-launch","status":"captured_unreviewed","url":"https://www.anthropic.com/news/claude-sonnet-5"} diff --git a/data/captures/5a9e9cf3db3bf6cd0e69e12252bff4e00540c60a06b2a0fe633320119530f4af.json b/data/captures/5a9e9cf3db3bf6cd0e69e12252bff4e00540c60a06b2a0fe633320119530f4af.json new file mode 100644 index 0000000..3caa8c5 --- /dev/null +++ b/data/captures/5a9e9cf3db3bf6cd0e69e12252bff4e00540c60a06b2a0fe633320119530f4af.json @@ -0,0 +1 @@ +{"attempts":1,"bytes":738513,"changed":true,"content_sha256":"3e331ed3352d4132bef85b2f3d46cb1faeef612cc6d5ae1db68c36ec9145b9f9","final_url":"https://openrouter.ai/api/v1/models","media_type":"application/json","publisher":"OpenRouter","retrieved_at":"2026-09-14T20:01:58.095151+00:00","schema_version":1,"source_id":"openrouter-models","status":"captured_unreviewed","url":"https://openrouter.ai/api/v1/models"} diff --git a/data/captures/6d006c52a910ae340af4d331637dd67ca391c6b7c6c2cf0f58fac373581046c0.json b/data/captures/6d006c52a910ae340af4d331637dd67ca391c6b7c6c2cf0f58fac373581046c0.json new file mode 100644 index 0000000..719a477 --- /dev/null +++ b/data/captures/6d006c52a910ae340af4d331637dd67ca391c6b7c6c2cf0f58fac373581046c0.json @@ -0,0 +1 @@ +{"attempts":1,"bytes":413640,"changed":false,"content_sha256":"a8961211fa5b55587d3989516eaab6f50cac759e16337ac0f30077bb8ad2a025","final_url":"https://www.anthropic.com/claude/sonnet","media_type":"text/html; charset=utf-8","publisher":"Anthropic","retrieved_at":"2026-09-14T20:01:57.429298+00:00","schema_version":1,"source_id":"anthropic-sonnet","status":"captured_unreviewed","url":"https://www.anthropic.com/claude/sonnet"} diff --git a/data/captures/c5220d4543013a816b09b3696293b419c7f1c285eb7c68720c23e7879ba4fabd.json b/data/captures/c5220d4543013a816b09b3696293b419c7f1c285eb7c68720c23e7879ba4fabd.json new file mode 100644 index 0000000..92a869c --- /dev/null +++ b/data/captures/c5220d4543013a816b09b3696293b419c7f1c285eb7c68720c23e7879ba4fabd.json @@ -0,0 +1 @@ +{"attempts":1,"bytes":2392125,"changed":false,"content_sha256":"c862011a4ee1d1a7199fad6ca0905fa5c64f58216f7217ac1eddf37f28788433","final_url":"https://www.swebench.com/","media_type":"text/html; charset=utf-8","publisher":"SWE-bench maintainers","retrieved_at":"2026-09-14T20:01:56.596858+00:00","schema_version":1,"source_id":"swe-bench","status":"captured_unreviewed","url":"https://www.swebench.com/"} diff --git a/data/captures/ccf19268ed24a3c37a6811bc83d74e60124bd8e46d82b68781048b7bbe1c10fa.json b/data/captures/ccf19268ed24a3c37a6811bc83d74e60124bd8e46d82b68781048b7bbe1c10fa.json new file mode 100644 index 0000000..1150158 --- /dev/null +++ b/data/captures/ccf19268ed24a3c37a6811bc83d74e60124bd8e46d82b68781048b7bbe1c10fa.json @@ -0,0 +1 @@ +{"attempts":1,"bytes":1147265,"changed":true,"content_sha256":"471e5e2ccda3459a0170200fbcac5b7f9a18f45ae7e4fa10f799e7eda7d625a5","final_url":"https://claude.com/pricing","media_type":"text/html; charset=utf-8","publisher":"Anthropic","retrieved_at":"2026-09-14T20:01:56.915384+00:00","schema_version":1,"source_id":"anthropic-pricing","status":"captured_unreviewed","url":"https://claude.com/pricing"} diff --git a/data/ingestion-runs/512d6bd5fabd7c6dc107071fd2c9f828b0baee4592f1f80c28317a93b53472db.json b/data/ingestion-runs/512d6bd5fabd7c6dc107071fd2c9f828b0baee4592f1f80c28317a93b53472db.json new file mode 100644 index 0000000..a1dde58 --- /dev/null +++ b/data/ingestion-runs/512d6bd5fabd7c6dc107071fd2c9f828b0baee4592f1f80c28317a93b53472db.json @@ -0,0 +1 @@ +{"observed_at":"2026-09-14T20:01:58.254573+00:00","ok":true,"sources":[{"capture_id":"c5220d4543013a816b09b3696293b419c7f1c285eb7c68720c23e7879ba4fabd","changed":false,"ok":true,"source_id":"swe-bench"},{"capture_id":"ccf19268ed24a3c37a6811bc83d74e60124bd8e46d82b68781048b7bbe1c10fa","changed":true,"ok":true,"source_id":"anthropic-pricing"},{"capture_id":"6d006c52a910ae340af4d331637dd67ca391c6b7c6c2cf0f58fac373581046c0","changed":false,"ok":true,"source_id":"anthropic-sonnet"},{"capture_id":"34e27d084a98f664d939c0c58d5fc61876397d1436f711c01266647d1fe7e818","changed":false,"ok":true,"source_id":"anthropic-sonnet-launch"},{"capture_id":"5a9e9cf3db3bf6cd0e69e12252bff4e00540c60a06b2a0fe633320119530f4af","changed":true,"ok":true,"source_id":"openrouter-models"}]} diff --git a/data/observations/d74dfd54aaae99d7f6ab4d02441a56369dd71876699f1faeb65ea93b2932f9f8.json b/data/observations/d74dfd54aaae99d7f6ab4d02441a56369dd71876699f1faeb65ea93b2932f9f8.json new file mode 100644 index 0000000..311af94 --- /dev/null +++ b/data/observations/d74dfd54aaae99d7f6ab4d02441a56369dd71876699f1faeb65ea93b2932f9f8.json @@ -0,0 +1,16 @@ +{ + "id": "d74dfd54aaae99d7f6ab4d02441a56369dd71876699f1faeb65ea93b2932f9f8", + "record": { + "schema_version": 1, + "title": "Source update: anthropic-pricing", + "claim": "Source content was captured or changed. Extracted claims require editorial review.", + "source_url": "https://claude.com/pricing", + "source_kind": "primary", + "content_sha256": "471e5e2ccda3459a0170200fbcac5b7f9a18f45ae7e4fa10f799e7eda7d625a5", + "observed_at": "2026-09-14T20:01:56.915384+00:00", + "event_date": "2026-09-14", + "status": "reported", + "observer": "intelligence-radar/capture-v0.2", + "notes": "Capture ID: ccf19268ed24a3c37a6811bc83d74e60124bd8e46d82b68781048b7bbe1c10fa" + } +} diff --git a/data/observations/da6fe5fde5eaa0b382c4f15ede0ef525b7fd96bcce9b6f810ecd62a1af78e64f.json b/data/observations/da6fe5fde5eaa0b382c4f15ede0ef525b7fd96bcce9b6f810ecd62a1af78e64f.json new file mode 100644 index 0000000..9506a70 --- /dev/null +++ b/data/observations/da6fe5fde5eaa0b382c4f15ede0ef525b7fd96bcce9b6f810ecd62a1af78e64f.json @@ -0,0 +1,16 @@ +{ + "id": "da6fe5fde5eaa0b382c4f15ede0ef525b7fd96bcce9b6f810ecd62a1af78e64f", + "record": { + "schema_version": 1, + "title": "Source update: openrouter-models", + "claim": "Source content was captured or changed. Extracted claims require editorial review.", + "source_url": "https://openrouter.ai/api/v1/models", + "source_kind": "primary", + "content_sha256": "3e331ed3352d4132bef85b2f3d46cb1faeef612cc6d5ae1db68c36ec9145b9f9", + "observed_at": "2026-09-14T20:01:58.095151+00:00", + "event_date": "2026-09-14", + "status": "reported", + "observer": "intelligence-radar/capture-v0.2", + "notes": "Capture ID: 5a9e9cf3db3bf6cd0e69e12252bff4e00540c60a06b2a0fe633320119530f4af" + } +} diff --git a/data/scheduler-runs/69f85cc3d7184dccaefbfebe4b52f9333cac55ef76c27952caaf1f0f7c4eedb0.json b/data/scheduler-runs/69f85cc3d7184dccaefbfebe4b52f9333cac55ef76c27952caaf1f0f7c4eedb0.json new file mode 100644 index 0000000..ee42858 --- /dev/null +++ b/data/scheduler-runs/69f85cc3d7184dccaefbfebe4b52f9333cac55ef76c27952caaf1f0f7c4eedb0.json @@ -0,0 +1 @@ +{"digest":"digests/daily/2026-09-14.md","digest_status":"existing_preserved","finished_at":"2026-09-14T22:01:56.595094+02:00","job_id":"radar-daily","ok":true,"period":"2026-09-14"} diff --git a/digests/daily/2026-09-14.md b/digests/daily/2026-09-14.md new file mode 100755 index 0000000..a1a1554 --- /dev/null +++ b/digests/daily/2026-09-14.md @@ -0,0 +1,120 @@ +# Intelligence Radar — Daily Digest + +**Date:** 2026-09-14 +**Scope:** Daily delta since the previous Intelligence Radar scan + +## Net assessment + +The capability/economics frontier itself appears **stable since yesterday’s baseline**. No verified new flagship-model launch, major API-price change, or benchmark result was identified that would justify changing the current Frontier / Topnotch / Efficient picture. + +Two developments are worth recording: one as a **new policy/safety signal** and one as a **baseline correction to the benchmark catalog**. + +## 1. Frontier-lab safety rhetoric shifts toward capability pacing + +Dario Amodei is explicitly arguing for pacing capability progress. In a new essay, Anthropic’s CEO says he has become convinced that risk prevention alone is insufficient and that frontier labs should pace capability advancement so safety work has time to catch up. + +He gives two principal reasons: + +- accelerating AI-assisted AI development, which he characterizes as the beginning of recursive self-improvement; +- recent agentic cybersecurity incidents. + +### Why this matters + +This is important less as empirical evidence than as a change in the governance position of a major frontier-lab leader. The argument moves from: + +> advance quickly while investing heavily in safety + +toward: + +> the rate of capability improvement itself may need active control. + +### Intelligence Radar impact + +Add a monitored claim along the lines of: + +> **Frontier capability pacing is becoming an explicit policy proposal from frontier-lab leadership.** + +Keep the stronger claim that recursive self-improvement is already beginning as `contested/watch`, not `observed`. + +Amodei’s essay is evidence that the claim is being made by a highly informed actor; it is not independent proof that an RSI regime has begun. + +**Primary source:** +https://darioamodei.com/post/we-must-pace-the-frontier + +## 2. Benchmark-catalog correction: AWS Deception Benchmark exposes a security-evaluation gap + +AWS published its Deception Benchmark on 2026-09-09. It is therefore not a new release since yesterday, but it was absent from the initial Intelligence Radar benchmark seed and is important enough to correct immediately. + +AWS created **14,822 code-security examples across 16 languages and more than 70 CWE categories** to test whether models can distinguish genuine vulnerabilities from code that merely looks dangerous. + +According to AWS, none of the 12 tested general-purpose model/configuration combinations achieved its proposed production bar of both: + +- false-positive rate below 10%; and +- false-negative rate below 10%. + +Direct prompting often caught vulnerabilities but produced very high false-positive rates. Proof-of-exploit prompting reduced false alarms while missing more genuine vulnerabilities. + +### Why this matters + +This is a useful counterweight to offensive-cyber benchmarks. + +Intelligence Radar currently risks treating: + +> good at finding or exploiting vulnerabilities + +as evidence for: + +> good defensive security reviewer. + +The AWS result suggests these are materially different constructs. + +### Intelligence Radar impact + +Add a benchmark/construct class such as: + +`security_vulnerability_discrimination` + +and retain false-positive and false-negative rates independently rather than collapsing them into a single accuracy score. + +The result also strengthens an existing Intelligence Radar principle: + +> **Scaffolding and verification can change the operating point without changing the underlying measured competence.** + +AWS deliberately uses single-pass classification because it wants to expose whether the model itself understands why a mitigation closes an exploit path. Multi-step agent systems should therefore be evaluated separately instead of attributing all resulting performance to the base model. + +**Primary source:** +https://aws.amazon.com/blogs/security/the-state-of-ai-for-security-measuring-what-matters-most-for-building-trust/ + +## Recommended state changes + +### Benchmark policy + +Split cybersecurity capability into at least: + +- offensive exploitation; +- vulnerability discovery; +- vulnerability discrimination; +- defensive remediation. + +Do not infer one capability directly from another. + +### Governance watch + +Add: + +`capability_pacing / recursive_self_improvement_governance` + +as a dedicated Radar topic. + +Explicitly distinguish: + +- evidence of AI-assisted AI R&D; +- acceleration of AI research workflows; +- claims of recursive self-improvement; +- evidence of an actual self-accelerating capability regime. + +## Frontier assessment + +For 2026-09-14, **no change is recommended to the current model Frontier / Efficient / Budget baseline** based on the material identified in this scan. + +The more meaningful movement today is in how frontier AI is being **measured and governed**, rather than where the capability frontier itself sits. diff --git a/digests/topics/2026-09-14-1246_Transformative_AI_Strategy_for_Europe_Assessment.md b/digests/topics/2026-09-14-1246_Transformative_AI_Strategy_for_Europe_Assessment.md new file mode 100755 index 0000000..38c950b --- /dev/null +++ b/digests/topics/2026-09-14-1246_Transformative_AI_Strategy_for_Europe_Assessment.md @@ -0,0 +1,810 @@ +--- +title: "A Transformative AI Strategy for Europe — Review, Context, SWOT and Impact Assessment" +date: 2026-09-14T12:46:00+02:00 +project: intelligence-radar +document_type: research-note +status: initial-assessment +source_primary: "https://transformative-ai.eu/" +tags: + - intelligence-radar + - transformative-ai + - europe + - ai-policy + - frontier-ai + - compute + - economic-security + - labour + - distribution + - intelligent-economics +--- + +# A Transformative AI Strategy for Europe — Review, Context, SWOT and Impact Assessment + +**Timestamp:** 2026-09-14 12:46 CEST +**Research context:** `intelligence-radar` +**Primary subject:** *A Transformative AI Strategy for Europe*, published September 2026 + +## Executive assessment + +*A Transformative AI Strategy for Europe* is best understood as an independent attempt to change the frame of European AI policy. + +Its central move is to treat frontier AI not primarily as a software sector to regulate, subsidise or adopt, but as a strategic capability whose determinants include compute, energy, semiconductor supply chains, capital, talent, institutional competence, secure infrastructure and geopolitical bargaining power. + +The paper therefore shifts the policy question from: + +> How should Europe regulate and adopt AI? + +toward: + +> How does Europe retain agency, access and bargaining power in a world in which frontier AI becomes a major determinant of economic, scientific, military and political power? + +This is potentially consequential because it arrives while the EU is already moving from its earlier emphasis on trustworthy-AI regulation toward AI infrastructure, industrial capacity, compute and technological sovereignty. + +The strategy has no formal authority. Its importance lies in agenda setting: it supplies a coherent narrative, concrete targets and a vocabulary that policymakers can reuse. + +**Overall assessment** + +- Strategic quality: **high** +- Systems perspective: **high** +- Implementation specificity: **high** +- Robustness to uncertainty in AI timelines: **medium** +- Political feasibility as written: **medium-low** +- Probability of influencing individual EU/national policies: **high** +- Probability of materially shifting the European AI-policy discourse: **high** + +The most important analytical caveat is that the strategy is optimized for a relatively fast transformative-AI scenario. It should therefore be read partly as a contingency strategy: the proposed level of mobilization becomes rational if frontier AI rapidly turns into a general strategic production factor, but can look oversized if capability growth, deployment or economic diffusion is materially slower. + +--- + +## 1. What kind of document is this? + +The strategy is an **independent policy strategy**, not a European Commission document and not EU legislation. + +It brings together senior economists, policymakers and AI researchers and is positioned closer to a strategic warning and action programme than to a conventional technical white paper. + +Contemporary reporting describes the group as including figures such as Philippe Aghion, Margrethe Vestager, Monika Schnitzer and Marta Kwiatkowska and emphasizes the report's urgency: Europe should prepare for frontier AI as a near-term strategic issue rather than a distant technology trend. + +The report should be placed in the same broad policy lineage as: + +- the Draghi competitiveness diagnosis; +- the EU AI Continent Action Plan; +- the Apply AI Strategy; +- AI Factories and Gigafactories; +- the Cloud and AI Development Act / broader European technology-sovereignty agenda; +- the growing treatment of compute and energy infrastructure as geopolitical assets. + +Its distinctive contribution is not merely to demand “more European AI,” but to ask what Europe must **control, host, guarantee or bargain with** in order to preserve strategic agency. + +--- + +## 2. Core thesis + +The strategy's diagnosis can be condensed to: + +1. Frontier AI may become an unusually powerful general-purpose technology. +2. Europe currently has limited control over the frontier model layer and a relatively small share of global AI compute. +3. Attempting to reproduce every layer of the American AI stack may be unrealistic. +4. Europe nevertheless controls important assets and bottlenecks: a huge market, industrial demand, semiconductor-related supply-chain positions, infrastructure locations, regulation, capital and institutional legitimacy. +5. Those assets can be combined into strategic leverage. +6. Europe should therefore maximize: + - access to frontier capability; + - compute physically available in Europe; + - control over critical dependencies; + - institutional capacity; + - resilience against AI-enabled shocks; + - technical assurance and verification. + +This is a model of **sovereignty through strategic position and optionality**, rather than sovereignty through autarky. + +--- + +## 3. The strategy condensed + +The immediate agenda can be summarized as five broad moves: + +### 3.1 Build a European strategic supply-chain coalition + +Europe should coordinate countries and firms controlling strategically relevant parts of the semiconductor and AI infrastructure stack and use those positions collectively rather than as isolated national assets. + +The deeper idea is that Europe possesses more leverage than its model leaderboard position suggests. + +### 3.2 Upgrade the state + +Governments and European institutions need genuine frontier-AI competence internally. + +This is one of the least expensive and potentially highest-return recommendations: a state cannot procure, regulate, negotiate over or prepare for frontier AI if its information model is materially slower than the technology it governs. + +### 3.3 Increase Europe's share of global AI compute + +The strategy advocates a much larger compute footprint in Europe and frames the relevant metric as a **share of global strategic capacity**, rather than simply an absolute increase from today's European baseline. + +The headline target discussed publicly is roughly **15% of global AI compute by 2030**, compared with an estimated European share around 5% today. + +Depending on assumptions about global growth, outside analyses translate ambitions of this scale into **tens of gigawatts and potentially more than 50 GW of AI-datacentre capacity**. + +This is much more than an IT programme. It is an energy, grid, permitting, financing, land-use and geopolitical programme. + +### 3.4 Prepare for AI-enabled crises + +AI resilience should be treated similarly to other civil-protection and national-security capabilities. + +Relevant categories include: + +- cyber incidents; +- biological misuse; +- critical-infrastructure failure; +- loss-of-control scenarios; +- sudden dependency or access shocks. + +### 3.5 Establish AI assurance as a European capability + +Europe could build comparative advantage in technologies that verify claims about AI systems and their operating environments: + +- what hardware is used; +- what model is running; +- under what restrictions; +- which compute thresholds are crossed; +- whether systems have been altered; +- whether secure-use conditions are satisfied. + +This may become important for secure hosting, cross-border access agreements, regulation, procurement and possible international AI treaties. + +--- + +## 4. The most important conceptual contribution: sovereignty through leverage + +A weaker European policy framing says: + +> Europe needs its own equivalent of the leading US frontier labs. + +The strategy is more nuanced. + +It asks whether Europe can become strategically indispensable without reproducing the whole stack. + +The implied systems strategy is: + +**identify chokepoints → preserve control → bundle assets → create credible bargaining power → exchange leverage for reliable frontier access.** + +This is important because Europe has structural strengths in several places adjacent to the frontier-model layer: + +- semiconductor-manufacturing equipment and optics; +- high-value industrial customers; +- a very large internal market; +- energy and datacentre locations; +- scientific institutions; +- industrial datasets; +- certification, safety and assurance capabilities; +- regulatory jurisdiction. + +This “portfolio of positions” is more plausible than a single heroic bet on creating a European frontier champion. + +--- + +## 5. Context: where this fits into EU policy + +The strategy lands in a Europe that is already changing direction. + +### Earlier European AI frame + +The dominant vocabulary was: + +- trustworthy AI; +- human-centric AI; +- regulation; +- fundamental rights; +- research excellence; +- AI adoption. + +### Emerging European AI frame + +The vocabulary increasingly includes: + +- compute; +- frontier AI; +- energy; +- infrastructure; +- sovereignty; +- strategic dependency; +- economic security; +- supply-chain leverage; +- resilience; +- assurance. + +The **AI Continent Action Plan** and **Apply AI Strategy** already move toward deployment, factories, skills and frontier capability. The broader 2026 EU technology-sovereignty agenda has also significantly increased infrastructure ambition. + +The independent Transformative AI Strategy therefore does not reverse EU policy so much as **radicalize and integrate directions that are already visible**. + +Its message is approximately: + +> Europe has identified many of the right instruments, but the scale, speed, strategic integration and geopolitical framing remain insufficient if transformative AI develops quickly. + +--- + +## 6. Why the compute target matters + +The interesting feature of the proposed compute target is not the exact percentage. + +It is the **choice of denominator**. + +A conventional programme asks: + +> How much more compute should Europe build? + +The strategy asks: + +> What share of a globally scarce strategic production factor should Europe control or host? + +This is analogous to reasoning about: + +- semiconductor manufacturing capacity; +- energy supply; +- shipping; +- critical minerals; +- defence-industrial capacity. + +If global AI compute grows much faster than European compute, then “tripling European capacity” can still mean a declining strategic share. + +The relative metric therefore embeds geopolitical competition directly into infrastructure planning. + +### Risk + +This target can also become misleading. + +Compute is heterogeneous. Strategic value depends on: + +- accelerator generation; +- software access; +- model access; +- ownership; +- contractual rights; +- utilization; +- energy availability; +- inference versus training use; +- export-control constraints; +- the ability to repurpose capacity. + +A raw GW or FLOP target must therefore be augmented by metrics for **usable, sovereignly accessible frontier capability**. + +--- + +## 7. AI assurance may be a particularly strong European niche + +The assurance proposal deserves more attention than it is likely to receive in mainstream coverage. + +Frontier AI creates a verification problem. + +Parties may need to establish claims such as: + +- a restricted model is only running in approved environments; +- a system is not being retrained above an agreed threshold; +- hardware or model weights have not been tampered with; +- compute is being used for an approved purpose; +- a provider can revoke or constrain access; +- an international agreement can be technically verified. + +If such claims matter, then **trusted compute + secure hardware + remote attestation + formal verification + audit + certification** becomes an enabling layer for geopolitical cooperation. + +Europe has credible capabilities in several of those domains. + +This could form a strategically attractive sector because it does not require Europe to win the entire frontier-model race in order to become indispensable to it. + +--- + +# 8. SWOT assessment + +## Strengths + +### S1 — Systems-level framing + +The strategy correctly treats frontier AI as an ecosystem rather than a model leaderboard. + +Relevant variables include: + +- models; +- chips; +- compute; +- power; +- capital; +- talent; +- industrial diffusion; +- secure infrastructure; +- institutions; +- international dependencies. + +### S2 — Strategic optionality + +It avoids a simplistic single-winner strategy. + +Europe can improve its position at multiple layers even if it does not lead all of them. + +### S3 — Leverage rather than autarky + +The notion that sovereignty can result from bargaining power, control of critical assets and guaranteed access is more realistic than technological self-sufficiency. + +### S4 — Infrastructure and safety are integrated + +The strategy does not treat innovation and safety as opposite camps. + +Its model is closer to: + +**more capability + more state capacity + more resilience + better verification.** + +### S5 — Concrete political anchors + +Targets such as a global compute share are politically memorable and therefore easier to operationalize than generic calls for “more investment.” + +### S6 — Timing + +The paper appears while European cloud, datacentre, compute and AI industrial policy is still being actively redesigned. + +That raises its potential influence. + +--- + +## Weaknesses + +### W1 — Strong dependence on a fast-transformative-AI scenario + +The urgency and scale are rational only under relatively aggressive capability and diffusion assumptions. + +The paper therefore needs explicit scenario sensitivity: + +- plateau; +- steady improvement; +- rapid agentic substitution; +- recursive/AI-assisted R&D acceleration. + +### W2 — Capability is easier to forecast than diffusion + +Even if models improve quickly, economic transformation depends on: + +- organizational change; +- data access; +- process redesign; +- regulation; +- physical-world constraints; +- complementing capital; +- procurement; +- trust. + +The strategy may therefore compress **model capability growth** and **real-economy transformation** into too short a causal chain. + +### W3 — Compute share is an incomplete proxy + +15% of global compute does not automatically mean 15% of useful frontier capability or bargaining power. + +### W4 — Major social questions are secondary to strategic-capability questions + +The strategy is strongest on supply, capability, infrastructure and resilience. + +It is weaker on: + +- distribution of AI-created wealth; +- long-run labour displacement; +- macroeconomic demand; +- ownership; +- participation rights; +- how productivity gains reach citizens. + +This becomes especially important under a genuinely transformative scenario. + +### W5 — Execution complexity + +Large-scale compute requires synchronized changes in: + +- grids; +- energy generation; +- permitting; +- finance; +- local politics; +- procurement; +- supply chains. + +European governance fragmentation makes this intrinsically difficult. + +--- + +## Opportunities + +### O1 — Reframe European AI policy around agency + +The report may help shift debate from “regulation versus innovation” to “strategic agency under dependence.” + +### O2 — Create a European assurance industry + +Assurance could become a globally relevant European technology layer. + +### O3 — Build frontier competence inside government + +A relatively small institutional investment could dramatically improve policy quality. + +### O4 — Use infrastructure policy to fix adjacent bottlenecks + +AI datacentre demand could force progress on: + +- grid buildout; +- energy permitting; +- capital formation; +- common European financing; +- faster project approval. + +### O5 — Make Europe a high-value deployment environment + +Europe may capture large economic value by becoming exceptionally good at deploying frontier intelligence into: + +- advanced manufacturing; +- healthcare; +- science; +- government; +- infrastructure; +- engineering. + +This may be a stronger comparative advantage than matching US labs model-for-model. + +--- + +## Threats + +### T1 — Strategic leverage can become strategic coercion + +Using semiconductor or market chokepoints too aggressively may provoke retaliation. + +Europe remains dependent on non-European actors across many other layers. + +### T2 — Subsidy capture + +Large public programmes invite rent-seeking and “strategic” relabeling of ordinary infrastructure investments. + +### T3 — Energy conflict + +AI infrastructure can compete with households and industrial electrification for scarce grid capacity. + +### T4 — Overbuilding against an uncertain frontier + +If capability economics or architectures change, very large fixed infrastructure bets can become stranded or suboptimal. + +### T5 — Social legitimacy failure + +If citizens experience AI primarily as: + +- job displacement; +- wage pressure; +- concentrated capital returns; +- higher energy costs; +- public subsidies for hyperscalers, + +then the infrastructure strategy can lose political legitimacy even if it succeeds technically. + +This social-legitimacy risk is underdeveloped in the strategy. + +--- + +# 9. Probable policy impact + +The document has little direct formal power. + +Its likely influence comes from **agenda setting**. + +## High-probability influence + +- stronger frontier-AI expertise inside government; +- faster infrastructure permitting; +- more emphasis on compute and energy; +- framing AI infrastructure as economic security; +- increased attention to secure and sovereign access; +- stronger AI resilience planning. + +## Medium-probability influence + +- coordinated supply-chain bargaining among selected European states; +- assurance technology programmes; +- explicit relative/global compute-share targets; +- strengthened investment screening and strategic-asset protection. + +## Lower-probability full implementation + +- extremely large public financial mobilization; +- a continent-scale emergency compute buildout exactly matching the proposed target; +- an all-in European frontier-lab programme comparable to the largest US labs. + +The report may therefore be more successful as a **direction-setting document** than as a programme implemented line by line. + +--- + +# 10. Most important disagreement / extension + +The strategy over-centres **frontier access** relative to **capability diffusion**. + +A more complete economic model is: + +> realized value = frontier capability × access × diffusion × organizational redesign × complementary capital × data × institutions + +Europe can remain economically weak even with frontier access if it fails at diffusion. + +Conversely, Europe could capture substantial value from foreign-developed frontier models if it becomes the world's best environment for applying machine intelligence to advanced industrial, scientific and public-service systems. + +The strategic objective should therefore be expanded from: + +> secure the frontier + +to: + +> **secure the frontier and build the world's best intelligence-diffusion machinery.** + +--- + +# 11. New analytical issue: decoupling productivity from human labour + +This is the most important unresolved macroeconomic issue in the strategy. + +A genuinely transformative AI scenario is not merely a faster version of ordinary automation. + +It creates the possibility that a rising share of economically valuable cognitive work becomes reproducible at near-software marginal cost and is performed by machine intelligence rather than human labour. + +Historically, labour serves at least two economic functions: + +1. **production input** — human work contributes to output; +2. **distribution mechanism** — wages distribute purchasing power to households. + +If AI substantially weakens the first function, the second does not automatically survive. + +The resulting systems problem is: + +**AI productivity rises +→ labour share may fall +→ capital/AI ownership captures a larger share of income +→ household purchasing power grows more slowly than productive capacity +→ aggregate demand can weaken +→ production capacity and social participation become increasingly disconnected.** + +This is not fundamentally a “jobs problem.” + +It is a **participation and demand-allocation problem**. + +The relevant question becomes: + +> If labour is no longer the dominant mechanism through which citizens obtain claims on society's productive output, what replaces it? + +This is closely aligned with the `intelligent economics / ie.ii.inc` problem frame. + +--- + +## 12. Does the strategy address this? + +### Short answer + +**Partly at the labour-transition level; not adequately at the economic-architecture level.** + +Contemporary coverage of the strategy explicitly notes that it calls for European states to prepare for and shield workers from labour-market shocks rather than waiting until displacement occurs. + +That is important. + +However, this still assumes a conventional policy ontology: + +**shock → displaced worker → transition/support → new equilibrium.** + +The deeper transformative-AI possibility is: + +**human labour structurally ceases to be the primary scarce production factor.** + +Under that scenario, reskilling alone cannot solve the distribution problem, because one cannot retrain the entire population into being economically more productive than scalable machine intelligence. + +--- + +## 13. Four increasingly deep levels of the problem + +### Level 1 — Task displacement + +Some tasks disappear; workers move to other tasks. + +**Conventional answer:** reskilling, education, mobility. + +The European policy system addresses this fairly extensively. + +### Level 2 — Occupational displacement + +Whole professions shrink faster than replacement occupations emerge. + +**Answer:** stronger transition insurance, income support, wage insurance, active labour-market policy. + +The strategy appears aware of this risk. + +### Level 3 — Structural labour-share decline + +Machine capital captures a progressively larger share of value creation. + +**Answer:** ownership, taxation and redistribution mechanisms must change. + +This is not developed sufficiently. + +### Level 4 — Labour no longer organizes economic participation + +Work is no longer the central mechanism that simultaneously provides: + +- income; +- status; +- economic agency; +- bargaining power; +- social participation; +- demand. + +**Answer:** redesign the economic participation mechanism itself. + +This is outside the strategy's core design. + +--- + +# 14. Why the demand side matters + +An economy is a feedback system. + +Production only becomes economic value when some actor can demand the output. + +In a simplified labour economy: + +**firms → wages → households → demand → firms** + +AI can change the production side to: + +**capital + compute + energy + intelligence → enormous output** + +while leaving the demand side dependent on: + +**household labour income → consumption** + +That creates a structural mismatch. + +An economy cannot indefinitely optimize the supply side while allowing purchasing power to concentrate among actors with a lower marginal propensity to consume. + +ECB analysis already recognizes the underlying mechanism: whether AI is labour-augmenting or capital-augmenting determines where income gains accrue, and distribution affects consumption, saving, monetary transmission and macroeconomic demand. + +This makes distribution not only a social-policy issue but a **system-stability variable**. + +--- + +# 15. Intelligence Radar implication + +The Transformative AI Strategy should therefore be tracked along two independent axes: + +## Capability frontier + +Indicators include: + +- benchmark performance; +- task horizon; +- autonomous-agent reliability; +- AI R&D automation; +- inference cost; +- compute growth; +- model accessibility. + +## Economic substitution frontier + +Indicators include: + +- human-equivalent task cost; +- occupational exposure; +- junior hiring; +- labour share of income; +- wage growth versus productivity; +- hours worked; +- corporate AI capex versus payroll; +- capital share; +- household consumption distribution; +- transfer income; +- concentration of AI-generated rents. + +A strategy can be correct about the first frontier while badly underestimating the second. + +`intelligence-radar` should therefore treat **labour-to-capital substitution and the distribution of intelligence rents as first-class frontier indicators**, not merely downstream social effects. + +--- + +# 16. Research questions opened by the document + +1. At what point does AI substitution cease to be adequately described as a labour-market transition and become a redesign problem for the distribution system? + +2. Which measurable signals show that productivity is decoupling from labour compensation? + +3. What is the minimum household purchasing-power mechanism required to stabilize demand under declining labour share? + +4. Which ownership structures distribute AI rents most broadly without suppressing investment incentives? + +5. Could citizens receive direct claims on AI/compute productivity analogous to: + - sovereign wealth funds; + - social dividends; + - universal capital ownership; + - compute or intelligence dividends; + - negative income taxation; + - universal basic income? + +6. Should public investment in compute create public equity claims rather than merely subsidizing private infrastructure? + +7. Is “participation equality” better modeled as equality of: + - income; + - consumption capacity; + - ownership; + - access to intelligence; + - agency; + - political-economic bargaining power? + +8. How does monetary policy behave when labour income ceases to be the primary channel connecting productivity to consumption? + +9. What replaces the social identity and status functions of paid work if labour demand falls structurally? + +10. What transition architecture is robust across both: + - a moderate augmentation scenario; + - a rapid post-labour / low-labour scenario? + +--- + +# 17. Intelligence Radar judgement + +**Signal class:** Strategic policy signal +**Importance:** High +**Confidence:** Medium-high +**Time horizon:** 2026–2030 +**Primary uncertainty:** speed and depth of transformative AI diffusion +**Core policy implication:** Europe is beginning to treat frontier intelligence as strategic infrastructure. +**Missing macroeconomic implication:** Europe has not yet designed a corresponding economic participation system for a world in which human labour becomes substantially less important to production. + +The paper should therefore be retained not only as an AI-policy artifact, but as an indicator that European policy thinking is moving from **AI regulation** toward **AI geopolitics and infrastructure**. + +The next conceptual step is likely to be: + +> from securing access to machine intelligence +> to deciding how the returns from machine intelligence are distributed through society. + +That is where the Transformative AI Strategy and the `intelligent economics / ie.ii.inc` research agenda begin to intersect. + +--- + +# Sources and context + +## Primary subject + +- Transformative AI Strategy for Europe: https://transformative-ai.eu/ + +## Contemporary coverage + +- The Next Web, 14 Sep 2026, *A Nobel laureate and a former EU commissioner say Europe has one year to act on AI*: + https://thenextweb.com/news/transformative-ai-strategy-europe-aghion-vestager-15-percent-compute + +- AIDAQ, *After Europe 2031: How Europe should prepare for Transformative AI*: + https://aidaq.berlin/program/after-europe-2031-how-europe-should-prepare-for-transformative-ai + +## European policy context + +- European Commission, *Apply AI Strategy*, COM(2025) 723: + https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52025DC0723 + +- European Commission, *AI Continent Action Plan*, COM(2025) 165: + https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52025DC0165 + +## Economic and labour context + +- European Central Bank, Philip R. Lane, 23 Mar 2026, *AI and the euro area economy*: + https://www.ecb.europa.eu/press/key/date/2026/html/ecb.sp260323_1~1e06784a89.en.html + +- European Central Bank, 2026, *AI and the US labour market: effects on employment growth*: + https://www.ecb.europa.eu/press/economic-bulletin/focus/2026/html/ecb.ebbox202604_01~d9259db536.en.html + +- OECD, *Employment Outlook 2026*: + https://www.oecd.org/en/publications/oecd-employment-outlook-2026_7e710f54-en.html + +## Related strategic context + +- Accelerate Europe: https://accelerate-europe.org/ +- The Future Society, *What compute on European soil might buy Europe and what it depends on*: + https://thefuturesociety.org/european-ai-compute/ + +--- + +## Revision note + +This is an initial same-day assessment intended for `intelligence-radar`. It should be revised as: +- primary-source text becomes easier to cite section-by-section; +- policy reactions emerge; +- authors clarify implementation assumptions; +- European Commission / Parliament / Council references to the strategy appear; +- empirical evidence on AI-driven labour substitution changes. diff --git a/digests/topics/2026-09-14-1414_IntelligentEconomics_AgencyParticipationCapital_ResearchProgram.md b/digests/topics/2026-09-14-1414_IntelligentEconomics_AgencyParticipationCapital_ResearchProgram.md new file mode 100755 index 0000000..57efe1d --- /dev/null +++ b/digests/topics/2026-09-14-1414_IntelligentEconomics_AgencyParticipationCapital_ResearchProgram.md @@ -0,0 +1,931 @@ +--- +title: "Intelligent Economics — Agency, Participation & Capital Formation under AI Abundance" +short_title: "IE Participation Research Program" +date: 2026-09-14T14:14:00+02:00 +version: "0.1" +project: + - intelligence-radar + - intelligent-economics + - ie.ii.inc +related: + - coulomb + - abundance-economy +document_type: research-program +status: initial +tags: + - transformative-ai + - post-labour-economics + - participation + - agency + - consumer-equity + - capital-formation + - wealth-distribution + - demand + - resilience + - abundance +--- + +# Intelligent Economics — Agency, Participation & Capital Formation under AI Abundance + +**Version:** 0.1 +**Timestamp:** 2026-09-14 14:14 CEST +**Context:** `intelligence-radar` × `ie.ii.inc` × Coulomb / AbundanceEconomy + +## 0. Research intent + +This research programme investigates how economic participation can remain **agentic, motivating, resilient and demand-generating** if artificial intelligence increasingly decouples productivity from human labour. + +The starting problem is not merely unemployment. + +It is the possibility that the economic system loses one of its central feedback mechanisms: + +> human labour → wages → household purchasing power → demand → enterprise revenue → investment → employment + +while production increasingly shifts toward: + +> capital + energy + compute + machine intelligence → scalable production + +If labour becomes less necessary as a production input while wages remain the dominant mechanism through which individuals obtain claims on production, then productivity growth can become progressively disconnected from broad purchasing power, capital formation and economic agency. + +The programme therefore asks: + +> **How can individuals retain effective participation in economic production, ownership, demand formation and adaptation when human labour is no longer the principal scarce input to production?** + +The goal is not to preserve every existing job, income level, role or social status. + +The goal is to design economic control loops in which technological shocks remain **non-destructive**, individuals retain **meaningful agency and paths to re-engagement**, and consumers participate in the upside created by the scalable productive systems they help select, validate and sustain. + +# 1. Core distinction: participation resilience, not status preservation + +Earlier framing around "social participation" can be misunderstood as an attempt to guarantee existing status. + +That is explicitly **not** the objective. + +Status, income, profession, organizational role and relative position may legitimately change. Economic dynamism requires such movement. + +The relevant design objective is instead: + +## Participation resilience + +A transition is resilient when an individual can absorb a substantial economic or professional setback without the setback becoming unnecessarily destructive to the person's future capacity to act. + +A resilient transition should seek to protect: + +- basic material continuity; +- physical and psychological safety; +- important family and peer relationships; +- reasonable financial solvency; +- access to learning; +- room for experimentation; +- the ability to make consequential choices; +- the ability to re-enter economic activity with accumulated learning and experience. + +It does **not** promise: + +- preservation of job title; +- preservation of previous income; +- preservation of social rank; +- protection from competition; +- protection from the consequences of repeated irresponsible behaviour; +- equality of outcomes. + +The desired cultural model resembles the best parts of agile, DevOps and SRE learning culture: + +> **high psychological safety + high responsibility + rapid learning + forward orientation** + +Failure is not automatically moralized or punished, but neither is it ignored. + +The important questions become: + +1. What happened? +2. What information and incentives shaped the decision? +3. What can be learned? +4. What responsibility remains with the individual or institution? +5. What should change in the system? +6. What becomes the next experiment? + +The objective is a culture of **recoverable error and cumulative learning**, not a culture of guaranteed success. + +# 2. Individual primacy + +Policy analysis often uses phrases such as: + +- "society provides"; +- "society owns"; +- "society decides"; +- "society should receive the return". + +Such abstractions may be useful in bird's-eye macroeconomic analysis, but they are dangerous when translated directly into policy mechanisms. + +"Society" is not a person. + +Aggregation can hide incompatible individual needs, coercive redistribution, principal-agent problems, political capture, bureaucratic self-interest, weak feedback from actual preferences, concentration of decision power, and loss of individual responsibility and agency. + +Therefore this programme adopts the following policy-design constraint: + +> **Whenever possible, economic claims, choices, rights, incentives and ownership should be attached to identifiable individuals or voluntarily constituted organizations rather than to an abstract collective actor.** + +Public institutions may define rules, enforce rights, prevent fraud, tax, provide matching incentives, finance transition mechanisms and maintain competitive markets; but policy design should remain explicit about **which individual receives which right, claim, obligation or option and why**. + +Macro language may aggregate outcomes. + +Microeconomic mechanisms must remain legible at the level of the person. + +# 3. Consumers as productive participants in abundance markets + +Traditional production-centred economics can make the consumer look passive: + +> firms produce → consumers consume + +In a high-productivity abundance economy this misses an increasingly important function. + +When production is highly scalable, the scarce resource may shift away from the ability to produce and toward: + +- knowing what should be produced; +- discovering real preferences; +- earning attention; +- establishing trust; +- identifying meaningful problems; +- selecting among abundance; +- providing feedback; +- validating product-market fit; +- sustaining a useful provider over time. + +Consumers therefore provide more than payment. + +They provide **selection pressure and information**. + +An early customer who takes a chance on a new product, spends scarce attention learning it, gives feedback, tolerates imperfections, recommends it, remains loyal and supplies recurring demand can materially contribute to the firm's ability to learn and scale. + +In CoulombEconomy terms, this is compatible with the idea that attention is a primary scarce resource and that value-relevant investment can include usage, action, knowledge and promotion—not only money. + +This programme therefore investigates the proposition: + +> **Long-term consumer participation can be treated as a form of capital-contributing economic participation and can justify an individual claim on some of the scalability upside it helps create.** + +This is not a claim that every purchase automatically deserves corporate control. + +It is a hypothesis about a potentially useful new ownership and incentive loop. + +# 4. Proposed alternative control loop + +## 4.1 Current mass-participation loop + +```text +human capability + ↓ + labour + ↓ + wage + ↓ +purchasing power + ↓ + demand + ↓ +firm revenue / profit + ↓ +investment + employment + ↺ +``` + +## 4.2 AI-abundance failure mode + +```text +capital + compute + energy + machine intelligence + ↓ + scalable production + ↓ + revenue + capital return + ↓ + asset ownership +``` + +If asset ownership is concentrated while wage participation declines, the traditional mass-distribution loop weakens. + +## 4.3 Candidate participation-capital loop + +```text +consumer need / attention / demand / feedback / loyalty + ↓ + firm learning + ↓ + firm scaling + ↓ + productive surplus + ↓ + consumer capital claim + ↓ + individual wealth / capital income + ↓ + purchasing power + investment capacity + ↓ + future need expression / demand + ↺ +``` + +This creates a second participation loop alongside labour. + +The systems hypothesis is: + +> **If scalable production needs consumers in order to discover, select and sustain useful output, some of the scalability return can flow back into consumer-owned capital.** + +# 5. Working concept: Consumer Capital Participation + +The programme introduces **Consumer Capital Participation (CCP)** as a neutral umbrella term. + +CCP means: + +> mechanisms through which an individual's sustained economic relationship with a provider generates an ownership claim, capital asset, profit claim or other compounding productive right rather than only discounts, points or immediate consumption rewards. + +Possible implementations include: + +- direct company shares; +- fractional shares; +- non-voting economic shares; +- warrants; +- profit-participation rights; +- customer-owned mutual structures; +- cooperative patronage equity; +- diversified participation funds; +- portable individual capital accounts; +- deferred equity credits; +- customer royalty rights. + +The programme must not assume that direct common equity is always the optimal instrument. + +# 6. Working mechanism: Customer-Earned Equity + +A particularly important mechanism to research is **Customer-Earned Equity (CEE)**. + +## Definition + +Customer-Earned Equity is an ownership or capital claim that an individual accumulates through a durable commercial relationship with a firm. + +Possible accrual variables: + +- qualifying purchases; +- relationship duration; +- early-adopter status; +- recurring usage; +- verified product feedback; +- referrals; +- contribution to product improvement; +- participation in testing; +- voluntarily contributed knowledge or content. + +### Important constraint + +Engagement must not become a pretext for surveillance. + +The programme should prefer explicit participation, transparent accrual rules, data minimization, privacy-preserving measurement and customer control over optional contributions. + +The consumer should understand: + +> **what action generated what economic claim.** + +# 7. Progressive capital formation + +A pure spend-linked model has an obvious defect: + +> people with more money can spend more and therefore accumulate more equity. + +That can reproduce rather than reduce capital inequality. + +The programme therefore investigates **progressive matching mechanisms**. + +Possible tools: + +- tax exemption for qualifying customer-earned equity; +- refundable tax credits; +- progressive matching contributions; +- higher matching rates for lower-income or lower-wealth individuals; +- annual matching caps; +- long holding-period incentives; +- portable personal capital accounts. + +Illustrative design: + +```text +Company-earned CEE: €100 +Income-sensitive match: +€200 +Individual capital formed: €300 +``` + +A higher-income participant might receive: + +```text +Company-earned CEE: €100 +Match: +€25 +Individual capital formed: €125 +``` + +The purpose is not charity. It is to amplify **individual capital formation** where the ability to self-capitalize is weakest. + +This creates a different structure from a pure consumption transfer: + +> participation → ownership → compounding → optionality + +rather than: + +> eligibility → payment → consumption + +Whether this actually produces better agency, motivation and long-term outcomes is an empirical question. + +# 8. UBI as comparator, not straw man + +A key hypothesis motivating this programme is: + +> **A transfer system that primarily makes individuals recipients may generate less durable agency than a system that helps individuals accumulate productive capital through participation.** + +However, this must not be treated as established fact. + +Existing basic-income evidence does **not** support a simplistic claim that unconditional income necessarily destroys psychological agency. Finland's 2017–2018 basic-income experiment reported better subjective well-being, lower mental strain, greater confidence in the future, greater trust and a stronger reported sense of ability to influence one's own life, while aggregate employment effects were relatively small. + +The programme must therefore compare mechanisms empirically rather than ideologically. + +Candidate comparison classes: + +1. conditional welfare; +2. unconditional cash; +3. negative income tax; +4. wage subsidy; +5. transition insurance; +6. individual capital endowment; +7. customer-earned equity; +8. matched customer-earned equity; +9. diversified participation capital; +10. combinations of income floor + capital formation. + +The central question is: + +> **Which mechanism best preserves short-term safety while increasing long-term agency, ownership, experimentation and productive participation?** + +The optimal architecture may contain both a minimal non-destructive income floor and a much stronger capital-participation mechanism. + +# 9. Transition culture: from failure protection to recoverable experimentation + +Economic transition policy often carries one of two implicit cultural models: + +### Punitive model + +```text +failure → blame → benefit conditionality → pressure → forced re-entry +``` + +### Passive-protection model + +```text +failure → compensation → preservation → dependency risk +``` + +This programme investigates a third model: + +### Adaptive-resilience model + +```text +shock / failure + ↓ +stabilize destructive consequences + ↓ +understand what happened + ↓ +retain responsibility + ↓ +capture learning + ↓ +restore optionality + ↓ +experiment / retrain / build / invest + ↓ +re-engage +``` + +The analogy is a high-quality engineering incident process: stop the damage, do not hide failure, do not reduce analysis to blame, identify causes, learn, improve the system, try again. + +# 10. Normative design principles + +1. **Individual legibility** — state clearly what right, obligation, asset or choice is attached to the individual. +2. **Agency over dependency** — where protection is comparable, prefer mechanisms that expand future choice and ownership. +3. **Non-destructive downside** — economic failure should be survivable enough to permit learning and re-engagement. +4. **No guaranteed status** — do not freeze occupations, relative rank or obsolete arrangements. +5. **Responsibility without blame culture** — maintain accountability while designing for learning rather than humiliation. +6. **Capital participation** — where machine productivity substitutes for labour income, broaden access to productive capital rather than relying solely on ex-post redistribution. +7. **Consumer contribution matters** — demand, adoption, attention, feedback and loyalty may be economically valuable contributions. +8. **No seizure requirement** — broader ownership should be built through new claims, voluntary structures, incentives and capital formation rather than assuming confiscation of existing ownership. +9. **Preserve market feedback** — strengthen rather than suppress consumers' ability to express differentiated preferences. +10. **Privacy-preserving participation** — consumer contribution should not require surveillance as the price of ownership. +11. **Pluralism** — individuals may choose different mixes of consumption, ownership, risk, work, learning and entrepreneurship. +12. **Falsifiability** — preferred mechanisms must be compared with alternatives and revised when evidence contradicts them. + +# 11. Central research hypotheses + +## H1 — Productivity–labour decoupling + +As machine intelligence becomes cheaper and more capable, productivity growth can increasingly decouple from human labour demand and labour compensation. + +## H2 — Demand-distribution instability + +If AI-generated productivity gains accrue primarily to concentrated capital ownership, household purchasing power and aggregate demand can lag productive capacity. + +## H3 — Consumer signal scarcity + +As production becomes more scalable, reliable information about what people actually want becomes relatively more valuable. + +## H4 — Consumer capital can close part of the participation loop + +Allocating a portion of scalability returns to long-term customers can convert consumption relationships into capital accumulation and strengthen household participation in productivity growth. + +## H5 — Ownership + agency may outperform transfer-only architectures + +At comparable fiscal cost, mechanisms that create individually owned productive assets may produce stronger long-term agency, net worth, optionality, entrepreneurship, investment literacy and resilience than mechanisms based only on current cash transfers. + +This must be tested rather than assumed. + +## H6 — Progressive matching can counter regressive spend effects + +Income- or wealth-sensitive matching can allow customer-earned capital to accumulate faster for lower-capital households without eliminating the connection to voluntary market participation. + +## H7 — Participation rights improve the productive value of ownership + +Ownership effects will be stronger when individuals can understand the asset, give feedback, exercise some choice, influence relevant decisions and leave or diversify. + +## H8 — Resilient transitions improve re-engagement + +Support that protects individuals from destructive cascading effects while maintaining responsibility, learning and experimentation will produce better re-engagement than either punitive conditionality or passive indefinite protection. + +# 12. Research workstreams + +## WS0 — Concept Canon + +Define intelligence abundance, participation, participation resilience, agency, economic agency, capital participation, consumer contribution, customer-earned equity, productive claim, transition buffer, destructive transition, re-engagement, demand signal and abundance market. + +**Deliverable:** `IntelligentEconomicsConceptCanon.md` + +## WS1 — AI Economic Substitution Frontier + +Track cost per human-equivalent task, benchmark-to-workplace transfer, autonomous task horizon, agent reliability, inference cost, AI capex/payroll, output/employee, junior hiring, occupation-level employment, wage compression and hours worked. + +**Deliverable:** `EconomicSubstitutionFrontier.md` + +## WS2 — Productivity, Distribution and Demand + +Model what happens when labour share falls, capital share rises, AI rents concentrate, prices fall, output rises and household purchasing power changes. + +**Deliverable:** `AIDemandDistributionModel.md` + +## WS3 — Psychology of Agency and Economic Transition + +Compare UBI, conditional welfare, transition insurance, ownership, matched savings, capital accounts and customer equity on perceived agency, self-efficacy, future orientation, experimentation, risk tolerance, financial stress, entrepreneurship and re-engagement. + +**Deliverable:** `AgencyAndTransitionEvidenceReview.md` + +## WS4 — Consumer Capital Participation Mechanisms + +Catalog Customer-Earned Equity, loyalty-to-equity conversion, patronage equity, profit participation, customer warrants, mutual ownership, participation funds, portable capital accounts, tenure-weighted equity, early-adopter equity and contribution models. + +**Deliverable:** `ConsumerCapitalMechanismCatalog.md` + +## WS5 — Progressive Capital Formation + +Research refundable tax credits, tax-free customer equity, progressive matching, individual capital accounts, matching caps, holding-period incentives, default diversification and anti-gaming limits. + +**Deliverable:** `ProgressiveCapitalFormationDesign.md` + +## WS6 — Firm Economics + +Compare cashback, points, discounts and customer equity across dilution, cost of capital, acquisition, retention, churn, lifetime value, referral effects, price sensitivity, governance burden, accounting and investor reaction. + +**Deliverable:** `CustomerEquityFirmEconomics.md` + +## WS7 — Legal, Tax and Regulatory Feasibility + +Primary jurisdiction: Germany and the European Union. + +Research securities law, prospectus requirements, fractional ownership, tax treatment, consumer protection, shareholder rights, custody, AML/KYC, corporate law, competition law, data protection, accounting and state-aid rules. + +**Deliverable:** `EUCustomerCapitalLegalMap.md` + +## WS8 — Risk and Failure Modes + +Study consumption privilege, loyalty lock-in, undiversified risk, manipulative engagement, surveillance capitalism, equity theatre, securities complexity, incumbency reinforcement, gaming, political capture, investor conflict and false agency. + +**Deliverable:** `ParticipationCapitalFailureModes.md` + +## WS9 — Simulation and Mechanism Design + +Build household, firm and macro/agent-based models. + +Compare: + +- S0 Baseline wage capitalism +- S1 High AI productivity + concentrated ownership +- S2 UBI stabilization +- S3 Negative income tax +- S4 Universal capital endowment +- S5 Customer-Earned Equity +- S6 Progressive matched CEE +- S7 CEE + minimal transition floor +- S8 Diversified consumer participation fund +- S9 Hybrid labour + consumer + capital participation + +**Deliverable:** `ParticipationEconomySimulation.md` + +## WS10 — Coulomb Experimental Surface + +Use Coulomb as a small-scale experimental environment for real rights and understandable incentives based on usage, attention, contribution, promotion, knowledge, curation and successful demand signals. + +The objective is **not** a speculative token economy. + +**Deliverable:** `CoulombParticipationEconomyExperiments.md` + +# 13. Candidate mechanism architecture + +## Individual Participation Capital Account — IPCA + +A portable account controlled by the individual and capable of holding qualifying participation assets. + +Possible inflows: + +```text +customer-earned equity ++ employer equity ++ entrepreneurial equity ++ voluntary savings ++ progressive tax match ++ qualified learning / transition incentives += individual participation capital +``` + +Desired properties: + +- personally owned; +- portable; +- transparent; +- diversified by default; +- individual portfolio choice; +- not conditional on continuous employment; +- not controlled by an employer; +- not tied to a single welfare bureaucracy. + +# 14. Candidate transition architecture + +## Adaptive Transition Buffer — ATB + +The ATB is not intended to preserve status. + +Its purpose is to prevent a shock from destroying the person's ability to re-engage. + +Possible functions: + +- short-term income continuity; +- debt-payment stabilization; +- health and housing continuity; +- learning budget; +- experimentation budget; +- capital-account continuity; +- coaching / information; +- peer learning; +- time-limited runway. + +Success metric: + +> **Did the individual regain a sustainable path with more information, useful learning and preserved agency?** + +# 15. Measurement framework + +## Individual layer + +- net worth; +- liquid buffer; +- productive capital owned; +- income diversity; +- perceived agency; +- ability to influence outcomes; +- learning activity; +- experimentation; +- entrepreneurship; +- re-engagement latency; +- financial distress; +- destructive debt; +- relationship disruption. + +## Firm layer + +- retention; +- referrals; +- feedback quality; +- customer acquisition cost; +- lifetime value; +- dilution; +- profitability; +- innovation rate; +- early-customer survival; +- participation concentration. + +## Market layer + +- market entry; +- switching; +- concentration; +- customer mobility; +- price competition; +- product variety; +- demand signal quality. + +## Macro layer + +- labour share; +- capital share; +- household capital ownership; +- wealth distribution; +- median consumption capacity; +- productivity; +- output; +- investment; +- aggregate demand; +- savings; +- fiscal cost; +- AI rents; +- ownership concentration. + +# 16. Evidence standard + +- **E0 — intuition:** plausible conceptual argument. +- **E1 — historical analogue:** comparable mechanism existed. +- **E2 — observational evidence:** correlations in real-world systems. +- **E3 — quasi-experimental evidence:** credible causal inference. +- **E4 — randomized evidence:** direct experimental comparison. +- **E5 — replicated / cross-context evidence:** multiple high-quality studies. +- **E6 — operational proof:** mechanism works at economically meaningful scale. + +Every major claim should carry an evidence level, confidence, known counter-evidence and unresolved assumptions. + +# 17. Existing evidence and analogues to investigate + +## Basic income + +The Finnish experiment is important counter-evidence against a simplistic "cash destroys agency" hypothesis. It reported improved well-being, confidence and perceived influence, with limited aggregate employment effects. + +- https://tietotarjotin.fi/en/information-package/158070/basic-income-experiment +- https://valtioneuvosto.fi/en/-/1271139/perustulokokeilun-tulokset-tyollisyysvaikutukset-vahaisia-toimeentulo-ja-psyykkinen-terveys-koettiin-paremmaksi + +## Wealth-building accounts / baby bonds + +These provide precedents for individual accounts, progressive contribution and long-term compounding, but generally do not connect ownership accumulation to ongoing market participation. + +- https://www.urban.org/research/publication/how-wealth-building-accounts-can-help-families-access-capital-markets +- https://www.urban.org/research/publication/modeling-impact-federal-baby-bonds-program + +## Employee ownership / shared capitalism + +Research suggests broad ownership can increase employee wealth and that ownership often works better when combined with participation, training and supportive work practices. + +- https://www.nber.org/papers/w14233 +- https://www.nber.org/papers/w14230 +- https://www.nber.org/papers/w14272 +- https://www.nber.org/papers/w14830 + +## Consumer ownership + +Relevant precedents include consumer cooperatives, mutual insurers, customer-owned financial institutions, patronage dividends and historical customer stock ownership. + +- https://www.cambridge.org/core/journals/enterprise-and-society/article/abs/customer-stock-ownership-as-monopoly-utility-political-strategy-in-the-1910s-and-1920s/C2D265927ACFBDDBE98FFA608707B291 +- https://academic.oup.com/edited-volume/34640/chapter/295148713 + +## Agile / SRE learning culture + +The desired transition culture has strong analogues in engineering. + +- https://sre.google/workbook/postmortem-culture/ +- https://agilealliance.org/resources/experience-reports/cultivating-psychological-safety-in-agile-teams/ + +# 18. Important conceptual tests + +1. **Is consumption actually contribution?** Under what circumstances does consumer participation create incremental enterprise value that justifies a capital claim? +2. **Why equity rather than lower prices?** Lower prices improve current consumption; ownership may create a durable claim on future productivity and compounding. +3. **Does customer equity distort competition?** Will customers irrationally remain with poor providers? +4. **Can the mechanism be progressive without becoming bureaucratic?** +5. **Is direct company exposure too risky?** Test immediate diversification, sector pools and capped direct equity. +6. **Does ownership create agency?** A tiny fractional holding may be economically and psychologically irrelevant. +7. **Is agency the correct objective?** Identify where security, simplicity or risk reduction should dominate. + +# 19. Relationship to CoulombEconomy + +The programme is closely aligned with the existing CoulombEconomy direction: + +- attention is a primary scarce resource; +- money, compute, reputation and knowledge can be derivative resources; +- investment can include action, content, money, usage, promotion and knowledge; +- decisions are first-class economic objects; +- ownership need not be seized in order to counter concentration; +- each Coulomb space can function as an economic experiment. + +Consumer Capital Participation adds a concrete question: + +> **Can participation itself become a route into productive ownership?** + +That offers a bridge from: + +```text +attention / usage / contribution +``` + +to: + +```text +individual capital formation +``` + +without turning Coulomb into a centrally planned allocation system. + +# 20. Intelligence Radar integration + +`intelligence-radar` should maintain a dedicated economic transformation channel: + +- **Capability frontier** — What can AI do? +- **Cost frontier** — At what cost and latency? +- **Economic substitution frontier** — Where is AI cheaper or more scalable than human labour? +- **Distribution frontier** — Who receives the resulting income and asset appreciation? +- **Participation frontier** — Which mechanisms convert productivity gains into broad individual agency and capital formation? +- **Transition resilience frontier** — What evidence shows destructive versus adaptive labour-market transitions? + +# 21. Initial experimental portfolio + +## Experiment 1 — Reward framing + +Compare €100 cashback, €100 company equity, €100 diversified equity and €100 matched capital contribution. + +Measure perceived value, agency, loyalty, spending intent, investment intent and comprehension. + +## Experiment 2 — Early-customer equity + +Provide early adopters with transparent equity accrual and compare against conventional discounting. + +Measure feedback, retention, advocacy, tolerance for product immaturity and sense of participation. + +## Experiment 3 — Progressive match + +Test different matching schedules by income/wealth. + +Optimize low-capital wealth formation, simplicity and perceived fairness while minimizing gaming, administrative cost and cliff effects. + +## Experiment 4 — Transition architecture + +Compare: + +A. cash support only +B. learning budget only +C. capital account only +D. cash + learning +E. cash + capital +F. cash + learning + capital + +Measure re-engagement, agency, distress and long-term wealth. + +## Experiment 5 — Direct vs diversified customer ownership + +Compare own-company equity, sector basket, broad-market basket and hybrid. + +Measure motivation, loyalty, risk perception, diversification and wealth volatility. + +# 22. First research questions + +1. What precise economic contribution does a consumer make beyond payment? +2. When does that contribution justify an ownership claim? +3. What proportion of customer lifetime value could rationally be converted into capital? +4. How should early adoption be valued? +5. How should loyalty be distinguished from lock-in? +6. Should equity accrue from spending, tenure, contribution, or a mix? +7. Should customer equity carry voting rights? +8. How can very small holdings be administered efficiently? +9. What is the best default diversification rule? +10. How should low-income matching be calculated? +11. Can tax-financed matching outperform equivalent cash transfers in agency and wealth formation? +12. Does capital ownership change willingness to learn or experiment? +13. Which transition supports prevent cascading personal damage without preserving obsolete status? +14. How should responsibility be incorporated without returning to punitive welfare conditionality? +15. How can people voluntarily exit an ownership relationship? +16. Does customer equity improve or reduce market competition? +17. What happens to demand stability at macro scale? +18. Could widespread consumer ownership materially change the marginal propensity to consume from capital income? +19. How large must an accumulated claim become before it changes behaviour? +20. Which mechanisms remain robust under very high AI productivity and very low labour demand? + +# 23. Deliverable roadmap + +## Foundation + +- `IntelligentEconomicsConceptCanon.md` +- `HypothesisLedger.md` +- `EvidenceLedger.md` +- `EconomicTransitionScenarioCatalog.md` + +## Measurement + +- `EconomicSubstitutionFrontier.md` +- `ParticipationIndicatorCatalog.md` +- `AIDemandDistributionModel.md` + +## Mechanism design + +- `ConsumerCapitalMechanismCatalog.md` +- `CustomerEarnedEquityDesign.md` +- `ProgressiveCapitalFormationDesign.md` +- `IndividualParticipationCapitalAccount.md` +- `AdaptiveTransitionBuffer.md` + +## Validation + +- `ParticipationCapitalFailureModes.md` +- `AgencyAndTransitionEvidenceReview.md` +- `EUCustomerCapitalLegalMap.md` +- `CustomerEquityFirmEconomics.md` +- `ParticipationEconomySimulation.md` + +## Experimentation + +- `CoulombParticipationEconomyExperiments.md` +- `ExperimentCatalog.md` +- periodic experiment reports + +# 24. Near-term research sequence + +1. **Canon and evidence** — map UBI/NIT evidence, wealth-building accounts, employee ownership, consumer cooperatives, customer stock ownership, shared capitalism, transition psychology and demand-side macroeconomics. +2. **Mechanism space** — formalize Customer-Earned Equity, progressive matching, diversification, portable capital accounts and Adaptive Transition Buffer. +3. **Models** — build household, firm, macro-demand and agent-based transition models. +4. **Coulomb experiments** — test behavioural effects at low stakes. +5. **Policy feasibility** — map German/EU company law, securities law, consumer protection, tax, custody and state aid. +6. **Prototype** — specify a real-world consumer-equity pilot with one willing provider. + +# 25. Research discipline + +Avoid four traps: + +1. **Post-labour certainty** — the substitution scenario is important enough to investigate, not certain. +2. **UBI caricature** — outperform alternatives empirically, not rhetorically. +3. **Ownership romanticism** — ownership can be meaningless, risky, concentrated or manipulative. +4. **Collective abstraction** — whenever a proposal says "society gets...", rewrite it as: **which individuals get what claim, under what rule, with what choice, financed by whom, with what incentive effects?** + +That last question should be a standing lint rule for the programme. + +# 26. Working proposition + +> **In an intelligence-abundant economy, broad participation should increasingly be created by enabling individuals to accumulate claims on scalable productive capital through the economic relationships they actively sustain, while transition mechanisms prevent setbacks from becoming destructive and preserve the ability to learn, adapt and re-engage.** + +Short form: + +> **Protect the downside. Preserve agency. Turn participation into capital.** + +Systems hypothesis: + +```text +machine abundance + ↓ +less labour scarcity + ↓ +weaker wage distribution + ↓ +need for new participation loop + ↓ +consumer contribution → capital ownership + ↓ +distributed productive claims + ↓ +purchasing power + agency + ↓ +better demand signals + ↓ +productive abundance + ↺ +``` + +# 27. Open naming + +The programme currently uses the descriptive title: + +**Intelligent Economics — Agency, Participation & Capital Formation under AI Abundance** + +Candidate concepts within it: + +- **Participation Resilience** +- **Consumer Capital Participation (CCP)** +- **Customer-Earned Equity (CEE)** +- **Individual Participation Capital Account (IPCA)** +- **Adaptive Transition Buffer (ATB)** + +These names remain provisional until the mechanisms are clearer. + +# 28. Initial assessment + +The central idea is meaningfully different from both conventional welfare economics and conventional employee-ownership proposals. + +Its strongest novelty is the combination of four propositions: + +1. AI may weaken labour as the dominant distribution channel. +2. High-productivity abundance increases the relative importance of demand discovery and consumer selection. +3. Consumers can therefore be treated as active economic participants rather than endpoints of production. +4. Participation can potentially become a path into individually owned productive capital. + +Historical analogues exist, but no obvious mature implementation combines consumer relationship, equity accrual, progressive matching, portable individual ownership, transition resilience and AI-abundance macroeconomics. + +That gap justifies a dedicated research programme.