{"name":"models","site":"https://aimodelrace.com","snapshot":"2026-10-19","notice":"Illustrative placeholder data. Methods: https://aimodelrace.com/methodology","data":[{"slug":"claude-opus-5-5","name":"Claude Opus 5.5","lab":"Anthropic","family":"Claude Opus","kind":"llm","status":"active","releaseDate":"2026-09-22","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":4,"output":20,"unit":"per-1m-tokens"},"speed":96,"latency":4.4,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Fable-class capability at Opus prices: the default pick for serious agentic and coding work.","primaryRace":"overall","primaryRank":1,"url":"https://aimodelrace.com/models/claude-opus-5-5"},{"slug":"claude-fable-5-1","name":"Claude Fable 5.1","lab":"Anthropic","family":"Claude Fable","kind":"llm","status":"active","releaseDate":"2026-08-20","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":10,"output":50,"unit":"per-1m-tokens"},"speed":54,"latency":8.8,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Anthropic's Mythos-class top tier for the longest, hardest problems where cost is secondary.","primaryRace":"overall","primaryRank":4,"url":"https://aimodelrace.com/models/claude-fable-5-1"},{"slug":"claude-sonnet-5-5","name":"Claude Sonnet 5.5","lab":"Anthropic","family":"Claude Sonnet","kind":"llm","status":"active","releaseDate":"2026-09-28","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":2,"output":10,"unit":"per-1m-tokens"},"speed":142,"latency":2.6,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"The workhorse: faster and cheaper than Opus, and it edges ahead on agentic coding.","primaryRace":"overall","primaryRank":6,"url":"https://aimodelrace.com/models/claude-sonnet-5-5"},{"slug":"claude-haiku-5-5","name":"Claude Haiku 5.5","lab":"Anthropic","family":"Claude Haiku","kind":"llm","status":"active","releaseDate":"2026-10-07","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":0.1,"output":0.5,"unit":"per-1m-tokens"},"speed":262,"latency":0.7,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Tiny price, startling computer-use skill: the sub-agent and high-volume specialist.","primaryRace":"overall","primaryRank":14,"url":"https://aimodelrace.com/models/claude-haiku-5-5"},{"slug":"claude-opus-5","name":"Claude Opus 5","lab":"Anthropic","family":"Claude Opus","kind":"llm","status":"legacy","releaseDate":"2026-07-24","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":5,"output":25,"unit":"per-1m-tokens"},"speed":72,"latency":5.5,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"The first Claude 5 Opus: a big jump in computer use and abstract reasoning.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/claude-opus-5"},{"slug":"claude-fable-5","name":"Claude Fable 5","lab":"Anthropic","family":"Claude Fable","kind":"llm","status":"legacy","releaseDate":"2026-06-09","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":10,"output":50,"unit":"per-1m-tokens"},"speed":48,"latency":10,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"The first publicly available Mythos-class model, shipped behind new safety classifiers.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/claude-fable-5"},{"slug":"claude-opus-4-8","name":"Claude Opus 4.8","lab":"Anthropic","family":"Claude Opus","kind":"llm","status":"legacy","releaseDate":"2026-05-14","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":64000,"pricing":{"input":5,"output":25,"unit":"per-1m-tokens"},"speed":66,"latency":5,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Refined Opus 4 generation; later the safety fallback model for Fable 5.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/claude-opus-4-8"},{"slug":"claude-opus-4-6","name":"Claude Opus 4.6","lab":"Anthropic","family":"Claude Opus","kind":"llm","status":"legacy","releaseDate":"2026-02-05","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":64000,"pricing":{"input":5,"output":25,"unit":"per-1m-tokens"},"speed":62,"latency":5,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Brought a 1M-token context and best-in-class long-context recall to Opus.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/claude-opus-4-6"},{"slug":"claude-opus-4-5","name":"Claude Opus 4.5","lab":"Anthropic","family":"Claude Opus","kind":"llm","status":"legacy","releaseDate":"2025-11-24","retiredDate":null,"openWeights":false,"params":null,"contextWindow":200000,"maxOutput":64000,"pricing":{"input":5,"output":25,"unit":"per-1m-tokens"},"speed":58,"latency":5,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Cut Opus pricing by two thirds and took the SWE-bench crown in late 2025.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/claude-opus-4-5"},{"slug":"claude-sonnet-4-5","name":"Claude Sonnet 4.5","lab":"Anthropic","family":"Claude Sonnet","kind":"llm","status":"legacy","releaseDate":"2025-09-29","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":64000,"pricing":{"input":3,"output":15,"unit":"per-1m-tokens"},"speed":80,"latency":2.2,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Claimed the 'best coding model in the world' title in autumn 2025.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/claude-sonnet-4-5"},{"slug":"claude-haiku-4-5","name":"Claude Haiku 4.5","lab":"Anthropic","family":"Claude Haiku","kind":"llm","status":"legacy","releaseDate":"2025-10-15","retiredDate":null,"openWeights":false,"params":null,"contextWindow":200000,"maxOutput":64000,"pricing":{"input":1,"output":5,"unit":"per-1m-tokens"},"speed":150,"latency":0.8,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Sonnet 4-level coding at a third of the price.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/claude-haiku-4-5"},{"slug":"claude-opus-4-1","name":"Claude Opus 4.1","lab":"Anthropic","family":"Claude Opus","kind":"llm","status":"legacy","releaseDate":"2025-08-05","retiredDate":null,"openWeights":false,"params":null,"contextWindow":200000,"maxOutput":32000,"pricing":{"input":15,"output":75,"unit":"per-1m-tokens"},"speed":40,"latency":4,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Incremental Opus upgrade focused on real-world coding precision.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/claude-opus-4-1"},{"slug":"claude-opus-4","name":"Claude Opus 4","lab":"Anthropic","family":"Claude Opus","kind":"llm","status":"legacy","releaseDate":"2025-05-22","retiredDate":null,"openWeights":false,"params":null,"contextWindow":200000,"maxOutput":32000,"pricing":{"input":15,"output":75,"unit":"per-1m-tokens"},"speed":42,"latency":4,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Launched the Claude 4 era and Claude Code's mainstream moment.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/claude-opus-4"},{"slug":"claude-3-7-sonnet","name":"Claude 3.7 Sonnet","lab":"Anthropic","family":"Claude Sonnet","kind":"llm","status":"legacy","releaseDate":"2025-02-24","retiredDate":null,"openWeights":false,"params":null,"contextWindow":200000,"maxOutput":64000,"pricing":{"input":3,"output":15,"unit":"per-1m-tokens"},"speed":75,"latency":1.6,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"The first hybrid reasoning Claude and the model that made AI coding mainstream.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/claude-3-7-sonnet"},{"slug":"gpt-6-astra","name":"GPT-6 Astra","lab":"OpenAI","family":"GPT Astra","kind":"llm","status":"active","releaseDate":"2026-09-03","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1050000,"maxOutput":128000,"pricing":{"input":10,"output":50,"unit":"per-1m-tokens"},"speed":60,"latency":11.5,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"OpenAI's top tier: operates software like a person and leads the hardest math.","primaryRace":"overall","primaryRank":3,"url":"https://aimodelrace.com/models/gpt-6-astra"},{"slug":"gpt-6-1-sol","name":"GPT-6.1 Sol","lab":"OpenAI","family":"GPT Sol","kind":"llm","status":"active","releaseDate":"2026-09-29","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":2,"output":10,"unit":"per-1m-tokens"},"speed":121,"latency":4,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"The everyday frontier model: elite reasoning, great instructions-following, mid-tier price.","primaryRace":"overall","primaryRank":2,"url":"https://aimodelrace.com/models/gpt-6-1-sol"},{"slug":"gpt-6-sol","name":"GPT-6 Sol","lab":"OpenAI","family":"GPT Sol","kind":"llm","status":"legacy","releaseDate":"2026-09-22","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":2,"output":10,"unit":"per-1m-tokens"},"speed":115,"latency":4.2,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"GPT-6's paid default in ChatGPT, replaced a week later by 6.1 Sol in the API.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gpt-6-sol"},{"slug":"gpt-6-luna","name":"GPT-6 Luna","lab":"OpenAI","family":"GPT Luna","kind":"llm","status":"active","releaseDate":"2026-09-22","retiredDate":null,"openWeights":false,"params":null,"contextWindow":400000,"maxOutput":64000,"pricing":{"input":0.2,"output":1.6,"unit":"per-1m-tokens"},"speed":232,"latency":0.9,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"The free-tier GPT-6: fast, friendly and cheap enough for anything high-volume.","primaryRace":"overall","primaryRank":15,"url":"https://aimodelrace.com/models/gpt-6-luna"},{"slug":"gpt-5-6-sol","name":"GPT-5.6 Sol","lab":"OpenAI","family":"GPT Sol","kind":"llm","status":"legacy","releaseDate":"2026-06-25","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":4,"output":20,"unit":"per-1m-tokens"},"speed":100,"latency":5,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Introduced the Sol naming and served as OpenAI's flagship through summer 2026.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gpt-5-6-sol"},{"slug":"gpt-5-5","name":"GPT-5.5","lab":"OpenAI","family":"GPT-5","kind":"llm","status":"legacy","releaseDate":"2026-04-23","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":5,"output":30,"unit":"per-1m-tokens"},"speed":90,"latency":6,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Codename 'Spud': a fresh pre-train that pushed 1M context into the GPT-5 line.","primaryRace":"overall","primaryRank":9,"url":"https://aimodelrace.com/models/gpt-5-5"},{"slug":"gpt-5-4","name":"GPT-5.4","lab":"OpenAI","family":"GPT-5","kind":"llm","status":"legacy","releaseDate":"2026-03-05","retiredDate":null,"openWeights":false,"params":null,"contextWindow":400000,"maxOutput":128000,"pricing":{"input":2.5,"output":15,"unit":"per-1m-tokens"},"speed":85,"latency":6.5,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Merged the Codex coding line back into the main GPT-5 model.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gpt-5-4"},{"slug":"gpt-5-2","name":"GPT-5.2","lab":"OpenAI","family":"GPT-5","kind":"llm","status":"legacy","releaseDate":"2025-12-11","retiredDate":null,"openWeights":false,"params":null,"contextWindow":400000,"maxOutput":128000,"pricing":{"input":1.75,"output":14,"unit":"per-1m-tokens"},"speed":80,"latency":7,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"A fast-follow answer to Gemini 3 focused on professional knowledge work.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gpt-5-2"},{"slug":"gpt-5-1","name":"GPT-5.1","lab":"OpenAI","family":"GPT-5","kind":"llm","status":"legacy","releaseDate":"2025-11-12","retiredDate":null,"openWeights":false,"params":null,"contextWindow":400000,"maxOutput":128000,"pricing":{"input":1.25,"output":10,"unit":"per-1m-tokens"},"speed":75,"latency":7,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Warmer, more conversational GPT-5 with adaptive reasoning.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gpt-5-1"},{"slug":"gpt-5","name":"GPT-5","lab":"OpenAI","family":"GPT-5","kind":"llm","status":"legacy","releaseDate":"2025-08-07","retiredDate":null,"openWeights":false,"params":null,"contextWindow":400000,"maxOutput":128000,"pricing":{"input":1.25,"output":10,"unit":"per-1m-tokens"},"speed":70,"latency":9,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Unified OpenAI's chat and reasoning lines behind an automatic router.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gpt-5"},{"slug":"o3","name":"OpenAI o3","lab":"OpenAI","family":"OpenAI o-series","kind":"llm","status":"legacy","releaseDate":"2025-04-16","retiredDate":null,"openWeights":false,"params":null,"contextWindow":200000,"maxOutput":100000,"pricing":{"input":2,"output":8,"unit":"per-1m-tokens"},"speed":60,"latency":12,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"The reasoning model that learned to think with images and tools.","primaryRace":"overall","primaryRank":20,"url":"https://aimodelrace.com/models/o3"},{"slug":"gpt-4-5","name":"GPT-4.5","lab":"OpenAI","family":"GPT-4","kind":"llm","status":"legacy","releaseDate":"2025-02-27","retiredDate":"2025-07-14","openWeights":false,"params":null,"contextWindow":128000,"maxOutput":16000,"pricing":{"input":75,"output":150,"unit":"per-1m-tokens"},"speed":25,"latency":1.4,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"The giant non-reasoning model: superb taste and EQ, enormous price.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gpt-4-5"},{"slug":"o3-mini","name":"OpenAI o3-mini","lab":"OpenAI","family":"OpenAI o-series","kind":"llm","status":"legacy","releaseDate":"2025-01-31","retiredDate":null,"openWeights":false,"params":null,"contextWindow":200000,"maxOutput":100000,"pricing":{"input":1.1,"output":4.4,"unit":"per-1m-tokens"},"speed":150,"latency":8,"modalities":{"input":["text"],"output":["text"]},"tagline":"Cheap, fast reasoning that put chain-of-thought into everyday use.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/o3-mini"},{"slug":"gpt-4o","name":"GPT-4o","lab":"OpenAI","family":"GPT-4","kind":"llm","status":"legacy","releaseDate":"2024-05-13","retiredDate":"2026-02-16","openWeights":false,"params":null,"contextWindow":128000,"maxOutput":16000,"pricing":{"input":2.5,"output":10,"unit":"per-1m-tokens"},"speed":110,"latency":0.5,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"The omni model that defined ChatGPT for a year and a half.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gpt-4o"},{"slug":"gpt-oss-120b","name":"gpt-oss-120b","lab":"OpenAI","family":"gpt-oss","kind":"llm","status":"legacy","releaseDate":"2025-08-05","retiredDate":null,"openWeights":true,"params":"117B MoE (5.1B active)","contextWindow":131000,"maxOutput":131000,"pricing":{"input":0.15,"output":0.6,"unit":"per-1m-tokens"},"speed":300,"latency":0.62,"modalities":{"input":["text"],"output":["text"]},"tagline":"OpenAI's first open-weights model since GPT-2, built to run on a single GPU.","primaryRace":"overall","primaryRank":21,"url":"https://aimodelrace.com/models/gpt-oss-120b"},{"slug":"gemini-4-argon","name":"Gemini 4 Argon","lab":"Google DeepMind","family":"Gemini Pro","kind":"llm","status":"active","releaseDate":"2026-10-13","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":1000000,"pricing":{"input":2,"output":10,"unit":"per-1m-tokens"},"speed":132,"latency":5,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Google's multimodal monster: sees, watches and listens better than anything, and writes up to 1M tokens back.","primaryRace":"overall","primaryRank":5,"url":"https://aimodelrace.com/models/gemini-4-argon"},{"slug":"gemini-3-5-flash","name":"Gemini 3.5 Flash","lab":"Google DeepMind","family":"Gemini Flash","kind":"llm","status":"active","releaseDate":"2026-05-19","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":65000,"pricing":{"input":0.3,"output":2.5,"unit":"per-1m-tokens"},"speed":524,"latency":0.38,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Absurdly fast frontier-lite model with a free tier and strong agentic chops.","primaryRace":"overall","primaryRank":10,"url":"https://aimodelrace.com/models/gemini-3-5-flash"},{"slug":"gemini-3-1-pro","name":"Gemini 3.1 Pro","lab":"Google DeepMind","family":"Gemini Pro","kind":"llm","status":"legacy","releaseDate":"2026-02-19","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":65000,"pricing":{"input":2,"output":12,"unit":"per-1m-tokens"},"speed":110,"latency":7,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Doubled Gemini 3's reasoning scores and topped GPQA for months.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gemini-3-1-pro"},{"slug":"gemini-3-pro","name":"Gemini 3 Pro","lab":"Google DeepMind","family":"Gemini Pro","kind":"llm","status":"legacy","releaseDate":"2025-11-18","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":65000,"pricing":{"input":2,"output":12,"unit":"per-1m-tokens"},"speed":105,"latency":8,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Leapfrogged the field in November 2025 and topped LMArena on launch day.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gemini-3-pro"},{"slug":"gemini-3-flash","name":"Gemini 3 Flash","lab":"Google DeepMind","family":"Gemini Flash","kind":"llm","status":"legacy","releaseDate":"2025-12-17","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":65000,"pricing":{"input":0.5,"output":3,"unit":"per-1m-tokens"},"speed":220,"latency":0.8,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Pro-grade reasoning at Flash speed, made the default in the Gemini app.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gemini-3-flash"},{"slug":"gemini-2-5-pro","name":"Gemini 2.5 Pro","lab":"Google DeepMind","family":"Gemini Pro","kind":"llm","status":"legacy","releaseDate":"2025-03-25","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":65000,"pricing":{"input":1.25,"output":10,"unit":"per-1m-tokens"},"speed":140,"latency":10,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"The model that put Google back on top, with a then-unmatched 1M context.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gemini-2-5-pro"},{"slug":"gemini-2-0-flash","name":"Gemini 2.0 Flash","lab":"Google DeepMind","family":"Gemini Flash","kind":"llm","status":"legacy","releaseDate":"2025-02-05","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":8000,"pricing":{"input":0.1,"output":0.4,"unit":"per-1m-tokens"},"speed":250,"latency":0.4,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Fast, cheap multimodal model that powered Google's agent experiments.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/gemini-2-0-flash"},{"slug":"grok-4-7","name":"Grok 4.7","lab":"xAI","family":"Grok","kind":"llm","status":"active","releaseDate":"2026-09-21","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":2,"output":6,"unit":"per-1m-tokens"},"speed":112,"latency":4,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"xAI's best coder yet: elite math, live X knowledge, aggressive pricing.","primaryRace":"overall","primaryRank":7,"url":"https://aimodelrace.com/models/grok-4-7"},{"slug":"grok-4-6","name":"Grok 4.6","lab":"xAI","family":"Grok","kind":"llm","status":"legacy","releaseDate":"2026-08-12","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":2,"output":6,"unit":"per-1m-tokens"},"speed":106,"latency":4,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Monthly-cadence Grok release that set up 4.7.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/grok-4-6"},{"slug":"grok-4-1","name":"Grok 4.1","lab":"xAI","family":"Grok","kind":"llm","status":"legacy","releaseDate":"2025-11-17","retiredDate":null,"openWeights":false,"params":null,"contextWindow":2000000,"maxOutput":64000,"pricing":{"input":3,"output":15,"unit":"per-1m-tokens"},"speed":80,"latency":3,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Briefly topped LMArena with a big jump in emotional intelligence and style.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/grok-4-1"},{"slug":"grok-4","name":"Grok 4","lab":"xAI","family":"Grok","kind":"llm","status":"legacy","releaseDate":"2025-07-09","retiredDate":null,"openWeights":false,"params":null,"contextWindow":256000,"maxOutput":64000,"pricing":{"input":3,"output":15,"unit":"per-1m-tokens"},"speed":70,"latency":9,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Trained with RL at pre-training scale; set records on HLE and ARC-AGI-2 in mid-2025.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/grok-4"},{"slug":"grok-3","name":"Grok 3","lab":"xAI","family":"Grok","kind":"llm","status":"legacy","releaseDate":"2025-02-17","retiredDate":null,"openWeights":false,"params":null,"contextWindow":131000,"maxOutput":16000,"pricing":{"input":3,"output":15,"unit":"per-1m-tokens"},"speed":80,"latency":0.7,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Colossus-trained model that made xAI a frontier contender.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/grok-3"},{"slug":"muse-wonder","name":"Muse Wonder","lab":"Meta Superintelligence Labs","family":"Muse","kind":"llm","status":"active","releaseDate":"2026-10-12","retiredDate":null,"openWeights":false,"params":null,"contextWindow":1000000,"maxOutput":128000,"pricing":{"input":4,"output":20,"unit":"per-1m-tokens"},"speed":90,"latency":5,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"Meta's first true frontier model, with six reasoning levels and a memory system.","primaryRace":"overall","primaryRank":8,"url":"https://aimodelrace.com/models/muse-wonder"},{"slug":"muse-spark-open","name":"Muse Spark Open","lab":"Meta Superintelligence Labs","family":"Muse Open","kind":"llm","status":"active","releaseDate":"2026-10-15","retiredDate":null,"openWeights":true,"params":"~816B MoE (~42B active)","contextWindow":1000000,"maxOutput":64000,"pricing":{"input":0.6,"output":2.4,"unit":"per-1m-tokens"},"speed":125,"latency":1.4,"modalities":{"input":["text","image","audio","video"],"output":["text"]},"tagline":"The most capable US open-weights model: multimodal in, 1M context, runs on your own hardware.","primaryRace":"overall","primaryRank":12,"url":"https://aimodelrace.com/models/muse-spark-open"},{"slug":"llama-4-maverick","name":"Llama 4 Maverick","lab":"Meta Superintelligence Labs","family":"Llama","kind":"llm","status":"legacy","releaseDate":"2025-04-05","retiredDate":null,"openWeights":true,"params":"400B MoE (17B active)","contextWindow":1000000,"maxOutput":16000,"pricing":{"input":0.2,"output":0.6,"unit":"per-1m-tokens"},"speed":160,"latency":0.55,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Meta's first natively multimodal MoE Llama; strong value, weaker reasoning.","primaryRace":"overall","primaryRank":22,"url":"https://aimodelrace.com/models/llama-4-maverick"},{"slug":"kimi-k3","name":"Kimi K3","lab":"Moonshot AI","family":"Kimi","kind":"llm","status":"active","releaseDate":"2026-07-16","retiredDate":null,"openWeights":true,"params":"2.8T MoE (open)","contextWindow":1000000,"maxOutput":128000,"pricing":{"input":3,"output":12,"unit":"per-1m-tokens"},"speed":70,"latency":3,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"A 2.8-trillion-parameter open model that developers prefer for front-end work.","primaryRace":"overall","primaryRank":11,"url":"https://aimodelrace.com/models/kimi-k3"},{"slug":"kimi-k2-thinking","name":"Kimi K2 Thinking","lab":"Moonshot AI","family":"Kimi","kind":"llm","status":"legacy","releaseDate":"2025-11-06","retiredDate":null,"openWeights":true,"params":"1T MoE (32B active)","contextWindow":256000,"maxOutput":64000,"pricing":{"input":0.6,"output":2.5,"unit":"per-1m-tokens"},"speed":50,"latency":6,"modalities":{"input":["text"],"output":["text"]},"tagline":"Open reasoning agent that could chain hundreds of tool calls.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/kimi-k2-thinking"},{"slug":"kimi-k2","name":"Kimi K2","lab":"Moonshot AI","family":"Kimi","kind":"llm","status":"legacy","releaseDate":"2025-07-11","retiredDate":null,"openWeights":true,"params":"1T MoE (32B active)","contextWindow":128000,"maxOutput":16000,"pricing":{"input":0.6,"output":2.5,"unit":"per-1m-tokens"},"speed":50,"latency":0.8,"modalities":{"input":["text"],"output":["text"]},"tagline":"The trillion-parameter open model that kicked off the 'Kimi moment'.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/kimi-k2"},{"slug":"deepseek-v4-pro","name":"DeepSeek V4 Pro","lab":"DeepSeek","family":"DeepSeek V","kind":"llm","status":"active","releaseDate":"2026-04-24","retiredDate":null,"openWeights":true,"params":"1.6T MoE (49B active)","contextWindow":1000000,"maxOutput":384000,"pricing":{"input":1.74,"output":3.48,"unit":"per-1m-tokens"},"speed":60,"latency":5,"modalities":{"input":["text"],"output":["text"]},"tagline":"Open 1.6T MoE with elite competitive programming at a tenth of closed-model prices.","primaryRace":"overall","primaryRank":16,"url":"https://aimodelrace.com/models/deepseek-v4-pro"},{"slug":"deepseek-v4-1-flash","name":"DeepSeek V4.1 Flash","lab":"DeepSeek","family":"DeepSeek Flash","kind":"llm","status":"active","releaseDate":"2026-10-17","retiredDate":null,"openWeights":true,"params":"552B MoE (8B active)","contextWindow":1000000,"maxOutput":128000,"pricing":{"input":0.3,"output":1.2,"unit":"per-1m-tokens"},"speed":182,"latency":1.2,"modalities":{"input":["text"],"output":["text"]},"tagline":"MIT-licensed speed demon: near-frontier coding for pennies, with built-in search.","primaryRace":"overall","primaryRank":19,"url":"https://aimodelrace.com/models/deepseek-v4-1-flash"},{"slug":"deepseek-v3-2","name":"DeepSeek V3.2","lab":"DeepSeek","family":"DeepSeek V","kind":"llm","status":"legacy","releaseDate":"2025-12-01","retiredDate":null,"openWeights":true,"params":"671B MoE (37B active)","contextWindow":128000,"maxOutput":64000,"pricing":{"input":0.28,"output":0.42,"unit":"per-1m-tokens"},"speed":40,"latency":3,"modalities":{"input":["text"],"output":["text"]},"tagline":"Sparse attention made long context nearly free; IMO-gold-level math variant.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/deepseek-v3-2"},{"slug":"deepseek-v3-1","name":"DeepSeek V3.1","lab":"DeepSeek","family":"DeepSeek V","kind":"llm","status":"legacy","releaseDate":"2025-08-21","retiredDate":null,"openWeights":true,"params":"671B MoE (37B active)","contextWindow":128000,"maxOutput":64000,"pricing":{"input":0.56,"output":1.68,"unit":"per-1m-tokens"},"speed":45,"latency":3,"modalities":{"input":["text"],"output":["text"]},"tagline":"Hybrid think/no-think model with much stronger agent skills.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/deepseek-v3-1"},{"slug":"deepseek-r1","name":"DeepSeek R1","lab":"DeepSeek","family":"DeepSeek V","kind":"llm","status":"legacy","releaseDate":"2025-01-20","retiredDate":null,"openWeights":true,"params":"671B MoE (37B active)","contextWindow":128000,"maxOutput":32000,"pricing":{"input":0.55,"output":2.19,"unit":"per-1m-tokens"},"speed":30,"latency":15,"modalities":{"input":["text"],"output":["text"]},"tagline":"The open reasoning model that shook markets in January 2025.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/deepseek-r1"},{"slug":"qwen3-6-max","name":"Qwen3.6-Max","lab":"Alibaba Qwen","family":"Qwen Max","kind":"llm","status":"preview","releaseDate":"2026-04-20","retiredDate":null,"openWeights":false,"params":null,"contextWindow":256000,"maxOutput":64000,"pricing":{"input":1.6,"output":6.4,"unit":"per-1m-tokens"},"speed":70,"latency":3,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Alibaba's proprietary flagship, tuned hard for coding agents and science code.","primaryRace":"overall","primaryRank":13,"url":"https://aimodelrace.com/models/qwen3-6-max"},{"slug":"qwen3-max","name":"Qwen3-Max","lab":"Alibaba Qwen","family":"Qwen Max","kind":"llm","status":"legacy","releaseDate":"2025-09-24","retiredDate":null,"openWeights":false,"params":null,"contextWindow":256000,"maxOutput":64000,"pricing":{"input":1.2,"output":6,"unit":"per-1m-tokens"},"speed":60,"latency":1.5,"modalities":{"input":["text"],"output":["text"]},"tagline":"Trillion-parameter Qwen that brought Alibaba into the top tier.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/qwen3-max"},{"slug":"qwen3-235b","name":"Qwen3 235B","lab":"Alibaba Qwen","family":"Qwen Max","kind":"llm","status":"legacy","releaseDate":"2025-04-29","retiredDate":null,"openWeights":true,"params":"235B MoE (22B active)","contextWindow":128000,"maxOutput":32000,"pricing":{"input":0.2,"output":0.6,"unit":"per-1m-tokens"},"speed":70,"latency":1,"modalities":{"input":["text"],"output":["text"]},"tagline":"Apache-licensed hybrid reasoning family that became the default open base model.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/qwen3-235b"},{"slug":"glm-5-3","name":"GLM-5.3","lab":"Z.ai (Zhipu)","family":"GLM","kind":"llm","status":"active","releaseDate":"2026-08-14","retiredDate":null,"openWeights":true,"params":"Open weights (GLM-5.2 base)","contextWindow":1000000,"maxOutput":128000,"pricing":{"input":1.4,"output":4.4,"unit":"per-1m-tokens"},"speed":76,"latency":2.5,"modalities":{"input":["text"],"output":["text"]},"tagline":"Post-training-only upgrade that turned GLM into a serious open terminal agent.","primaryRace":"overall","primaryRank":18,"url":"https://aimodelrace.com/models/glm-5-3"},{"slug":"glm-4-5","name":"GLM-4.5","lab":"Z.ai (Zhipu)","family":"GLM","kind":"llm","status":"legacy","releaseDate":"2025-07-28","retiredDate":null,"openWeights":true,"params":"355B MoE (32B active)","contextWindow":128000,"maxOutput":96000,"pricing":{"input":0.6,"output":2.2,"unit":"per-1m-tokens"},"speed":70,"latency":1.2,"modalities":{"input":["text"],"output":["text"]},"tagline":"MIT-licensed agentic model that made Chinese open weights a coding default.","primaryRace":"overall","primaryRank":null,"url":"https://aimodelrace.com/models/glm-4-5"},{"slug":"mistral-large-4","name":"Mistral Large 4","lab":"Mistral AI","family":"Mistral Large","kind":"llm","status":"preview","releaseDate":"2026-10-06","retiredDate":null,"openWeights":true,"params":"1T MoE (49B active)","contextWindow":256000,"maxOutput":64000,"pricing":{"input":2,"output":6,"unit":"per-1m-tokens"},"speed":108,"latency":1.2,"modalities":{"input":["text","image"],"output":["text"]},"tagline":"Europe's trillion-parameter answer: 160+ languages, weights promised to developers.","primaryRace":"overall","primaryRank":17,"url":"https://aimodelrace.com/models/mistral-large-4"},{"slug":"gpt-image-2","name":"GPT Image 2","lab":"OpenAI","family":"GPT Image","kind":"image","status":"active","releaseDate":"2026-05-12","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.08,"output":0.08,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"The prompt-adherence champion: does exactly what you describe, text included.","primaryRace":"image","primaryRank":1,"url":"https://aimodelrace.com/models/gpt-image-2"},{"slug":"nano-banana-2-1","name":"Nano Banana 2.1","lab":"Google DeepMind","family":"Nano Banana","kind":"image","status":"active","releaseDate":"2026-10-06","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.04,"output":0.04,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Conversational editing that keeps faces, products and scenes perfectly consistent.","primaryRace":"image","primaryRank":2,"url":"https://aimodelrace.com/models/nano-banana-2-1"},{"slug":"nano-banana-pro","name":"Nano Banana Pro","lab":"Google DeepMind","family":"Nano Banana","kind":"image","status":"legacy","releaseDate":"2025-11-20","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.134,"output":0.134,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Gemini 3 Pro Image: studio-grade realism, infographics and 4K output.","primaryRace":"image","primaryRank":null,"url":"https://aimodelrace.com/models/nano-banana-pro"},{"slug":"flux-3-klein","name":"FLUX.3 Klein","lab":"Black Forest Labs","family":"FLUX","kind":"image","status":"active","releaseDate":"2026-10-15","retiredDate":null,"openWeights":true,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.015,"output":0.015,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Two-second, open-weights generation with multi-reference editing and layout boxes.","primaryRace":"image","primaryRank":6,"url":"https://aimodelrace.com/models/flux-3-klein"},{"slug":"flux-2-pro","name":"FLUX.2 Pro","lab":"Black Forest Labs","family":"FLUX","kind":"image","status":"active","releaseDate":"2025-11-25","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.03,"output":0.03,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"4-megapixel photoreal workhorse with an open-weights Dev sibling.","primaryRace":"image","primaryRank":null,"url":"https://aimodelrace.com/models/flux-2-pro"},{"slug":"midjourney-v8","name":"Midjourney V8","lab":"Midjourney","family":"Midjourney","kind":"image","status":"active","releaseDate":"2026-03-18","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.05,"output":0.05,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Still the most beautiful images in the race, now at native 2K.","primaryRace":"image","primaryRank":5,"url":"https://aimodelrace.com/models/midjourney-v8"},{"slug":"seedream-5","name":"Seedream 5","lab":"ByteDance Seed","family":"Seedream","kind":"image","status":"active","releaseDate":"2026-06-02","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.03,"output":0.03,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Fast, cheap and accurate: the batch-production favourite.","primaryRace":"image","primaryRank":3,"url":"https://aimodelrace.com/models/seedream-5"},{"slug":"ideogram-4","name":"Ideogram 4.0","lab":"Ideogram","family":"Ideogram","kind":"image","status":"active","releaseDate":"2026-04-15","retiredDate":null,"openWeights":true,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.06,"output":0.06,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"The typography specialist, now downloadable as open weights.","primaryRace":"image","primaryRank":4,"url":"https://aimodelrace.com/models/ideogram-4"},{"slug":"qwen-image-2-1","name":"Qwen Image 2.1","lab":"Alibaba Qwen","family":"Qwen Image","kind":"image","status":"active","releaseDate":"2026-10-09","retiredDate":null,"openWeights":true,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.02,"output":0.02,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Open-weights bilingual text rendering and precise edits at bargain prices.","primaryRace":"image","primaryRank":7,"url":"https://aimodelrace.com/models/qwen-image-2-1"},{"slug":"grok-imagine","name":"Grok Imagine","lab":"xAI","family":"Grok Imagine","kind":"image","status":"active","releaseDate":"2026-08-04","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.02,"output":0.02,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Fast, playful generation with a dedicated anime mode, built into X.","primaryRace":"image","primaryRank":8,"url":"https://aimodelrace.com/models/grok-imagine"},{"slug":"recraft-v3","name":"Recraft V3","lab":"Recraft","family":"Recraft","kind":"image","status":"active","releaseDate":"2024-10-30","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.04,"output":0.04,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"The designer's model: brand styles, icons and editable vectors.","primaryRace":"image","primaryRank":10,"url":"https://aimodelrace.com/models/recraft-v3"},{"slug":"gpt-image-1","name":"GPT Image 1","lab":"OpenAI","family":"GPT Image","kind":"image","status":"legacy","releaseDate":"2025-04-23","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.07,"output":0.07,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Brought GPT-4o's viral image generation to the API.","primaryRace":"image","primaryRank":null,"url":"https://aimodelrace.com/models/gpt-image-1"},{"slug":"midjourney-v7","name":"Midjourney V7","lab":"Midjourney","family":"Midjourney","kind":"image","status":"legacy","releaseDate":"2025-04-03","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.05,"output":0.05,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Rebuilt from scratch with Draft Mode and personalization on by default.","primaryRace":"image","primaryRank":null,"url":"https://aimodelrace.com/models/midjourney-v7"},{"slug":"midjourney-v6-1","name":"Midjourney V6.1","lab":"Midjourney","family":"Midjourney","kind":"image","status":"legacy","releaseDate":"2024-07-30","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.05,"output":0.05,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"The 2024 aesthetic benchmark.","primaryRace":"image","primaryRank":null,"url":"https://aimodelrace.com/models/midjourney-v6-1"},{"slug":"imagen-4","name":"Imagen 4","lab":"Google DeepMind","family":"Imagen","kind":"image","status":"legacy","releaseDate":"2025-05-20","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.04,"output":0.04,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Google's photoreal text-to-image model with much better typography.","primaryRace":"image","primaryRank":9,"url":"https://aimodelrace.com/models/imagen-4"},{"slug":"flux-1-kontext","name":"FLUX.1 Kontext","lab":"Black Forest Labs","family":"FLUX","kind":"image","status":"legacy","releaseDate":"2025-05-29","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.04,"output":0.04,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"In-context editing that kept characters consistent across edits.","primaryRace":"image","primaryRank":null,"url":"https://aimodelrace.com/models/flux-1-kontext"},{"slug":"flux-1-1-pro","name":"FLUX1.1 Pro","lab":"Black Forest Labs","family":"FLUX","kind":"image","status":"legacy","releaseDate":"2024-10-02","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.04,"output":0.04,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Fast, photoreal and the open-community favourite of late 2024.","primaryRace":"image","primaryRank":null,"url":"https://aimodelrace.com/models/flux-1-1-pro"},{"slug":"nano-banana","name":"Nano Banana","lab":"Google DeepMind","family":"Nano Banana","kind":"image","status":"legacy","releaseDate":"2025-08-26","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.039,"output":0.039,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Gemini 2.5 Flash Image: the viral editing model of 2025.","primaryRace":"image","primaryRank":null,"url":"https://aimodelrace.com/models/nano-banana"},{"slug":"seedream-4","name":"Seedream 4.0","lab":"ByteDance Seed","family":"Seedream","kind":"image","status":"legacy","releaseDate":"2025-09-10","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.03,"output":0.03,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Unified generation and editing at up to 4K, briefly #1 on arenas.","primaryRace":"image","primaryRank":null,"url":"https://aimodelrace.com/models/seedream-4"},{"slug":"ideogram-3","name":"Ideogram 3.0","lab":"Ideogram","family":"Ideogram","kind":"image","status":"legacy","releaseDate":"2025-03-26","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.06,"output":0.06,"unit":"per-image"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["image"]},"tagline":"Best-in-class text rendering in early 2025.","primaryRace":"image","primaryRank":null,"url":"https://aimodelrace.com/models/ideogram-3"},{"slug":"veo-3-1","name":"Veo 3.1","lab":"Google DeepMind","family":"Veo","kind":"video","status":"active","releaseDate":"2025-10-15","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.15,"output":0.15,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video","audio"]},"tagline":"The all-rounder: native dialogue and sound, physical realism, up to 4K.","primaryRace":"video","primaryRank":3,"url":"https://aimodelrace.com/models/veo-3-1"},{"slug":"kling-3-0","name":"Kling 3.0","lab":"Kuaishou (Kling)","family":"Kling","kind":"video","status":"active","releaseDate":"2026-02-10","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.1,"output":0.1,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video","audio"]},"tagline":"Multi-shot storyboards with synced audio and the best hair, liquid and cloth.","primaryRace":"video","primaryRank":2,"url":"https://aimodelrace.com/models/kling-3-0"},{"slug":"seedance-2-0","name":"Seedance 2.0","lab":"ByteDance Seed","family":"Seedance","kind":"video","status":"active","releaseDate":"2026-03-12","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.022,"output":0.022,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video","audio"]},"tagline":"Image-to-video king with rock-solid subject consistency and dual-channel audio.","primaryRace":"video","primaryRank":1,"url":"https://aimodelrace.com/models/seedance-2-0"},{"slug":"runway-gen-4-5","name":"Runway Gen-4.5","lab":"Runway","family":"Runway Gen","kind":"video","status":"active","releaseDate":"2025-12-01","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.12,"output":0.12,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video"]},"tagline":"The director's tool: motion brushes, camera control and consistent characters.","primaryRace":"video","primaryRank":5,"url":"https://aimodelrace.com/models/runway-gen-4-5"},{"slug":"wan-2-7","name":"Wan 2.7","lab":"Alibaba Qwen","family":"Wan","kind":"video","status":"active","releaseDate":"2026-06-20","retiredDate":null,"openWeights":true,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.05,"output":0.05,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video","audio"]},"tagline":"Open-weights video with photoreal faces and native audio.","primaryRace":"video","primaryRank":4,"url":"https://aimodelrace.com/models/wan-2-7"},{"slug":"ltx-2-3","name":"LTX-2.3","lab":"Lightricks","family":"LTX","kind":"video","status":"active","releaseDate":"2026-05-06","retiredDate":null,"openWeights":true,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.02,"output":0.02,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video","audio"]},"tagline":"4K at 50fps with synced audio, running locally in ComfyUI.","primaryRace":"video","primaryRank":6,"url":"https://aimodelrace.com/models/ltx-2-3"},{"slug":"hailuo-2-3","name":"Hailuo 2.3","lab":"MiniMax","family":"Hailuo","kind":"video","status":"active","releaseDate":"2025-10-28","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.045,"output":0.045,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video"]},"tagline":"Believable skin and lighting on a budget.","primaryRace":"video","primaryRank":8,"url":"https://aimodelrace.com/models/hailuo-2-3"},{"slug":"luma-ray-3","name":"Luma Ray 3","lab":"Luma AI","family":"Ray","kind":"video","status":"active","releaseDate":"2025-09-18","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.08,"output":0.08,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video"]},"tagline":"The first native 16-bit HDR video model, with reasoning-driven drafts.","primaryRace":"video","primaryRank":9,"url":"https://aimodelrace.com/models/luma-ray-3"},{"slug":"grok-imagine-video","name":"Grok Imagine Video 1.5","lab":"xAI","family":"Grok Imagine","kind":"video","status":"active","releaseDate":"2026-10-08","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.05,"output":0.05,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video","audio"]},"tagline":"Quick, stylised clips with sound and a strong anime streak.","primaryRace":"video","primaryRank":7,"url":"https://aimodelrace.com/models/grok-imagine-video"},{"slug":"sora-2","name":"Sora 2","lab":"OpenAI","family":"Sora","kind":"video","status":"legacy","releaseDate":"2025-09-30","retiredDate":"2026-09-24","openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.1,"output":0.1,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video","audio"]},"tagline":"Photoreal video with dialogue and a social app; discontinued in September 2026.","primaryRace":"video","primaryRank":null,"url":"https://aimodelrace.com/models/sora-2"},{"slug":"veo-3","name":"Veo 3","lab":"Google DeepMind","family":"Veo","kind":"video","status":"legacy","releaseDate":"2025-05-20","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.4,"output":0.4,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video","audio"]},"tagline":"The model that ended the silent era of AI video.","primaryRace":"video","primaryRank":null,"url":"https://aimodelrace.com/models/veo-3"},{"slug":"veo-2","name":"Veo 2","lab":"Google DeepMind","family":"Veo","kind":"video","status":"legacy","releaseDate":"2024-12-16","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.35,"output":0.35,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video"]},"tagline":"Google's late-2024 leap in physical realism.","primaryRace":"video","primaryRank":null,"url":"https://aimodelrace.com/models/veo-2"},{"slug":"kling-2-5-turbo","name":"Kling 2.5 Turbo","lab":"Kuaishou (Kling)","family":"Kling","kind":"video","status":"legacy","releaseDate":"2025-09-23","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.07,"output":0.07,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video"]},"tagline":"Faster, cheaper Kling with much better motion.","primaryRace":"video","primaryRank":null,"url":"https://aimodelrace.com/models/kling-2-5-turbo"},{"slug":"kling-2-1","name":"Kling 2.1","lab":"Kuaishou (Kling)","family":"Kling","kind":"video","status":"legacy","releaseDate":"2025-05-29","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.09,"output":0.09,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video"]},"tagline":"Mid-2025 motion-quality leader.","primaryRace":"video","primaryRank":null,"url":"https://aimodelrace.com/models/kling-2-1"},{"slug":"kling-1-6","name":"Kling 1.6","lab":"Kuaishou (Kling)","family":"Kling","kind":"video","status":"legacy","releaseDate":"2024-12-19","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.09,"output":0.09,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video"]},"tagline":"The model that put Kling on the map.","primaryRace":"video","primaryRank":null,"url":"https://aimodelrace.com/models/kling-1-6"},{"slug":"runway-gen-4","name":"Runway Gen-4","lab":"Runway","family":"Runway Gen","kind":"video","status":"legacy","releaseDate":"2025-03-31","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.12,"output":0.12,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video"]},"tagline":"Introduced reference-based character consistency.","primaryRace":"video","primaryRank":null,"url":"https://aimodelrace.com/models/runway-gen-4"},{"slug":"seedance-1-0","name":"Seedance 1.0","lab":"ByteDance Seed","family":"Seedance","kind":"video","status":"legacy","releaseDate":"2025-06-11","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.03,"output":0.03,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video"]},"tagline":"Multi-shot generation that topped video arenas in mid-2025.","primaryRace":"video","primaryRank":null,"url":"https://aimodelrace.com/models/seedance-1-0"},{"slug":"hailuo-02","name":"Hailuo 02","lab":"MiniMax","family":"Hailuo","kind":"video","status":"legacy","releaseDate":"2025-06-18","retiredDate":null,"openWeights":false,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.045,"output":0.045,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video"]},"tagline":"Physics-heavy model famous for acrobatics clips.","primaryRace":"video","primaryRank":null,"url":"https://aimodelrace.com/models/hailuo-02"},{"slug":"wan-2-2","name":"Wan 2.2","lab":"Alibaba Qwen","family":"Wan","kind":"video","status":"legacy","releaseDate":"2025-07-28","retiredDate":null,"openWeights":true,"params":null,"contextWindow":null,"maxOutput":null,"pricing":{"input":0.03,"output":0.03,"unit":"per-second"},"speed":null,"latency":null,"modalities":{"input":["text","image"],"output":["video"]},"tagline":"Apache-licensed MoE video model for local generation.","primaryRace":"video","primaryRank":null,"url":"https://aimodelrace.com/models/wan-2-2"}]}