AI models

Compare models

Pick a benchmark that already has a score, filter, and mark up to four models.

Pick a benchmark that already has a score.
Filter by name, lab or weights.
Mark up to four models.
Token price does not close the cost of the finished job.

Benchmarks with a score

Case	What to measure	Reference model	Open alternative	Source
Text	GPQA Diamond	Gemini 4 Argon 96.4%	Kimi K3.1 Pro 94.2%	Epoch AI
Text	OTIS Mock AIME	Claude Opus 5.5 100%	Kimi K3.1 Pro 100%	Epoch AI
Text	FrontierMath (tiers 1 to 3)	Claude Fable 5.2 95.1%	Kimi K3.1 Pro 78.2%	Epoch AI
Text	SimpleQA Verified	Claude Fable 5.2 94%	Kimi K3.1 Pro 62.2%	Epoch AI
Code	SWE-bench Verified	Claude Opus 4.7 83.5%	GLM-5.2 78.7%	Epoch AI
Text	GPQA Diamond (OpenRouter)	Gemini 4 Argon 96.5%	Kimi K3.1 Pro 93.9%	OpenRouter Benchmarks
Agents	τ-bench Verified (airline)	Claude Fable 5 82%	DeepSeek V4 Flash 0423 81.7%	OpenRouter Benchmarks

GPQA Diamond

How to read each metric

Search modelLabWeights

GPQA Diamond
Model	%	Lab	Weights	Source	Compare models
Gemini 4 Argon	96.4%	Google	closed	Epoch AI	Gemini 4 Argon
GPT-6 Astra	95.8%	OpenAI	closed	Epoch AI	GPT-6 Astra
Claude Sonnet 5.5	95.6%	Anthropic	closed	Epoch AI	Claude Sonnet 5.5
GPT-6.1 Sol	95.4%	OpenAI	closed	Epoch AI	GPT-6.1 Sol
Gemini 3.8 Flash	95.4%	Google	closed	Epoch AI	Gemini 3.8 Flash
Gemini 3.7 Flash	94.8%	Google	closed	Epoch AI	Gemini 3.7 Flash
GPT-5.4 Pro	94.6%	OpenAI	closed	Epoch AI	GPT-5.4 Pro
Gemini 3.1 Pro	94.4%	Google	closed	Epoch AI	Gemini 3.1 Pro
GPT-6 Sol	94.3%	OpenAI	closed	Epoch AI	GPT-6 Sol
GPT-6.1 Terra	94.3%	OpenAI	closed	Epoch AI	GPT-6.1 Terra
Kimi K3.1 Pro	94.2%	Moonshot AI	open	Epoch AI	Kimi K3.1 Pro
Gemini 3.6 Flash	94.1%	Google	closed	Epoch AI	Gemini 3.6 Flash
GPT-5.5	94%	OpenAI	closed	Epoch AI	GPT-5.5
Grok 4.6	94%	xAI	closed	Epoch AI	Grok 4.6
Claude Fable 5.2	93.9%	Anthropic	closed	Epoch AI	Claude Fable 5.2
Claude Opus 5	93.9%	Anthropic	closed	Epoch AI	Claude Opus 5
GPT-5.5 Pro	93.9%	OpenAI	closed	Epoch AI	GPT-5.5 Pro
GPT-5.6 Sol	93.5%	OpenAI	closed	Epoch AI	GPT-5.6 Sol
Grok 4.5	93.4%	xAI	closed	Epoch AI	Grok 4.5
GPT-5.4	93.3%	OpenAI	closed	Epoch AI	GPT-5.4
GPT-5.6 Terra	93.3%	OpenAI	closed	Epoch AI	GPT-5.6 Terra
DeepSeek V4.1 Flash	93.2%	DeepSeek	open	Epoch AI	DeepSeek V4.1 Flash
Kimi K3	93.1%	Moonshot AI	open	Epoch AI	Kimi K3
Doubao Seed Evolving (Nov 2026)	93.1%	ByteDance	closed	Epoch AI	Doubao Seed Evolving (Nov 2026)
Gemini 3.5 Flash	92.8%	Google	closed	Epoch AI	Gemini 3.5 Flash
Qwen3.8 Max	92.7%	Alibaba	closed	Epoch AI	Qwen3.8 Max
Grok 4.7	92.7%	xAI	closed	Epoch AI	Grok 4.7
Gemini 3 Pro	92.6%	Google	closed	Epoch AI	Gemini 3 Pro
GLM-5.2	91.9%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.2
DeepSeek V4 Pro	91.7%	DeepSeek	open	Epoch AI	DeepSeek V4 Pro
GPT-5.6 Luna	91.6%	OpenAI	closed	Epoch AI	GPT-5.6 Luna
GPT-5.2	91.4%	OpenAI	closed	Epoch AI	GPT-5.2
Claude Opus 4.8	91%	Anthropic	closed	Epoch AI	Claude Opus 4.8
DeepSeek V4 Flash 0731	91%	DeepSeek	open	Epoch AI	DeepSeek V4 Flash 0731
GLM-5.3	90.9%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.3
MiniMax M3	90.9%	MiniMax	open	Epoch AI	MiniMax M3
Qwen3.7 Max	90.9%	Alibaba	closed	Epoch AI	Qwen3.7 Max
Kimi K2.6	90.8%	Moonshot AI	open	Epoch AI	Kimi K2.6
Claude Opus 5.5	90.6%	Anthropic	closed	Epoch AI	Claude Opus 5.5
GPT-6 Luna	90.5%	OpenAI	closed	Epoch AI	GPT-6 Luna
Claude Opus 4.6	90.5%	Anthropic	closed	Epoch AI	Claude Opus 4.6
Claude Sonnet 5	90.5%	Anthropic	closed	Epoch AI	Claude Sonnet 5
GLM-5.3-Flash	90.2%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.3-Flash
Mistral Large 4	90.2%	Mistral AI	open	Epoch AI	Mistral Large 4
Claude Opus 4.7	90.2%	Anthropic	closed	Epoch AI	Claude Opus 4.7
GLM 5.1	89.9%	Zhipu AI (Z.ai)	open	Epoch AI	GLM 5.1
Muse Spark	89.8%	Meta	closed	Epoch AI	Muse Spark
Claude Haiku 5.5	89.6%	Anthropic	closed	Epoch AI	Claude Haiku 5.5
Doubao Seed 2.1 Turbo	89.6%	ByteDance	closed	Epoch AI	Doubao Seed 2.1 Turbo
Gemini 3 Flash	89.4%	Google	closed	Epoch AI	Gemini 3 Flash
Grok 4.20	89.3%	xAI	closed	Epoch AI	Grok 4.20
Doubao Seed 2.1 Pro	89.3%	ByteDance	closed	Epoch AI	Doubao Seed 2.1 Pro
Grok 4.3 Beta	88.8%	xAI	closed	Epoch AI	Grok 4.3 Beta
Muse Spark Open	88.6%	Meta	open	Epoch AI	Muse Spark Open
Qwen3.6 Plus	88.4%	Alibaba	closed	Epoch AI	Qwen3.6 Plus
Doubao-Seed-2.0 Pro	88.2%	ByteDance	closed	Epoch AI	Doubao-Seed-2.0 Pro
Doubao Seed 2.0 Lite (Apr 2026)	88.1%	ByteDance	closed	Epoch AI	Doubao Seed 2.0 Lite (Apr 2026)
Kimi K2.7 Code	87.9%	Moonshot AI	open	Epoch AI	Kimi K2.7 Code
Qwen3.7 Plus	87.9%	Alibaba	closed	Epoch AI	Qwen3.7 Plus
GLM 5	87.8%	Zhipu AI (Z.ai)	open	Epoch AI	GLM 5
Kimi K2.5	87.6%	Moonshot AI	open	Epoch AI	Kimi K2.5
GPT-5.1	87.6%	OpenAI	closed	Epoch AI	GPT-5.1
Claude Sonnet 4.6	87.4%	Anthropic	closed	Epoch AI	Claude Sonnet 4.6
Qwen 3.6 Max (Preview)	87.4%	Alibaba	undisclosed	Epoch AI	Qwen 3.6 Max (Preview)
Grok 4	87%	xAI	closed	Epoch AI	Grok 4
GPT-5.4 Mini	86.9%	OpenAI	closed	Epoch AI	GPT-5.4 Mini
Qwen3.5 397B A17B	86.4%	Alibaba	open	Epoch AI	Qwen3.5 397B A17B
GPT-5	86.2%	OpenAI	closed	Epoch AI	GPT-5
Claude Opus 4.5	86%	Anthropic	closed	Epoch AI	Claude Opus 4.5
Claude Fable 5	85.9%	Anthropic	closed	Epoch AI	Claude Fable 5
Qwen3.6 27B	85.9%	Alibaba	open	Epoch AI	Qwen3.6 27B
Nemotron 3 Ultra	85.4%	NVIDIA	open	Epoch AI	Nemotron 3 Ultra
Gemini 2.5 Pro	85.3%	Google	closed	Epoch AI	Gemini 2.5 Pro
Qwen3.6 35B A3B	84.8%	Alibaba	open	Epoch AI	Qwen3.6 35B A3B
Qwen 3.5 Plus (hosted 397B-A17B)	84.8%	Alibaba	undisclosed	Epoch AI	Qwen 3.5 Plus (hosted 397B-A17B)
Kimi K2 Thinking	84.2%	Moonshot AI	open	Epoch AI	Kimi K2 Thinking
Qwen3.5-35B-A3B	83.5%	Alibaba	open	Epoch AI	Qwen3.5-35B-A3B
DeepSeek V3.2	83.4%	DeepSeek	open	Epoch AI	DeepSeek V3.2
Gemini 3.5 Flash Lite	83.3%	Google	closed	Epoch AI	Gemini 3.5 Flash Lite
Qwen3.6 Flash	83.3%	Alibaba	undisclosed	Epoch AI	Qwen3.6 Flash
GLM 4.7	83.3%	Zhipu AI (Z.ai)	open	Epoch AI	GLM 4.7
GPT-5.5 Instant	82.5%	OpenAI	closed	Epoch AI	GPT-5.5 Instant
Claude Sonnet 4.5	82.3%	Anthropic	closed	Epoch AI	Claude Sonnet 4.5
Qwen3.5-Flash	82.3%	Alibaba	undisclosed	Epoch AI	Qwen3.5-Flash
Qwen3.7 Flash	82.3%	Alibaba	closed	Epoch AI	Qwen3.7 Flash
Gemini 3.1 Flash Lite	81.8%	Google	closed	Epoch AI	Gemini 3.1 Flash Lite
o3	81.8%	OpenAI	closed	Epoch AI	o3
Qwen3-235B-A22B	80.1%	Alibaba	open	Epoch AI	Qwen3-235B-A22B
Claude 3.7 Sonnet	79.7%	Anthropic	closed	Epoch AI	Claude 3.7 Sonnet
o4-mini	79.6%	OpenAI	closed	Epoch AI	o4-mini
Claude Sonnet 4	79.2%	Anthropic	closed	Epoch AI	Claude Sonnet 4
Qwen3.5-9B	79%	Alibaba	open	Epoch AI	Qwen3.5-9B
GPT-5.4 Nano	78.5%	OpenAI	closed	Epoch AI	GPT-5.4 Nano
Doubao-Seed-1.6	77.6%	ByteDance	closed	Epoch AI	Doubao-Seed-1.6
Claude Opus 4.1	77.3%	Anthropic	closed	Epoch AI	Claude Opus 4.1
o3-mini	77%	OpenAI	closed	Epoch AI	o3-mini
o1	76.8%	OpenAI	closed	Epoch AI	o1
DeepSeek R1	76.3%	DeepSeek	open	Epoch AI	DeepSeek R1
Claude Opus 4	76.3%	Anthropic	closed	Epoch AI	Claude Opus 4
Grok-3 mini	76.3%	xAI	closed	Epoch AI	Grok-3 mini
gpt-oss-120b	75.8%	OpenAI	open	Epoch AI	gpt-oss-120b
Gemma 4 31B	75.8%	Google	open	Epoch AI	Gemma 4 31B
Grok 3	75.8%	xAI	closed	Epoch AI	Grok 3
GPT-5 Mini	75%	OpenAI	closed	Epoch AI	GPT-5 Mini
Gemma 4 26B A4B	73.2%	Google	open	Epoch AI	Gemma 4 26B A4B
Qwen3-Max	72.6%	Alibaba	closed	Epoch AI	Qwen3-Max
Claude Haiku 4.5	71.2%	Anthropic	closed	Epoch AI	Claude Haiku 4.5
Qwen3 30B A3B	70.1%	Alibaba	open	Epoch AI	Qwen3 30B A3B
GPT-5 Nano	69.4%	OpenAI	closed	Epoch AI	GPT-5 Nano
GPT-4.5	68.7%	OpenAI	closed	Epoch AI	GPT-4.5
DeepSeek V3	67.6%	DeepSeek	open	Epoch AI	DeepSeek V3
Llama 4 Maverick	67%	Meta	open	Epoch AI	Llama 4 Maverick
GPT-4.1	66.9%	OpenAI	closed	Epoch AI	GPT-4.1
GPT-4.1 Mini	65.8%	OpenAI	closed	Epoch AI	GPT-4.1 Mini
Qwen3 32B	65.7%	Alibaba	open	Epoch AI	Qwen3 32B
Gemini 2.0 Pro	65.7%	Google	closed	Epoch AI	Gemini 2.0 Pro
QWQ-Plus	65.4%	Alibaba	closed	Epoch AI	QWQ-Plus
QwQ-32B	65.3%	Alibaba	open	Epoch AI	QwQ-32B
DeepSeek-R1-Distill-Qwen-32B	64.1%	DeepSeek	open	Epoch AI	DeepSeek-R1-Distill-Qwen-32B
Gemini 2.0 Flash	64.1%	Google	closed	Epoch AI	Gemini 2.0 Flash
Qwen3 14B	63.8%	Alibaba	open	Epoch AI	Qwen3 14B
o1-mini	62.4%	OpenAI	closed	Epoch AI	o1-mini
gpt-oss-20b	60.8%	OpenAI	open	Epoch AI	gpt-oss-20b
Mistral Medium 3	59.5%	Mistral AI	closed	Epoch AI	Mistral Medium 3
Gemini 1.5 Pro	57.2%	Google	closed	Epoch AI	Gemini 1.5 Pro
Gemini 2.0 Flash Thinking	57.1%	Google	closed	Epoch AI	Gemini 2.0 Flash Thinking
Qwen3-8B	56.8%	Alibaba	open	Epoch AI	Qwen3-8B
DeepSeek V3	56.5%	DeepSeek	open	Epoch AI	DeepSeek V3
Phi 4	56.1%	Microsoft	open	Epoch AI	Phi 4
Magistral Small 1.0	56.1%	Mistral AI	open	Epoch AI	Magistral Small 1.0
Qwen2.5-Max	56.1%	Alibaba	closed	Epoch AI	Qwen2.5-Max
DeepSeek-R1-Distill-Llama-70B	55.7%	DeepSeek	open	Epoch AI	DeepSeek-R1-Distill-Llama-70B
Claude 3.5 Sonnet (October 2024)	55.3%	Anthropic	closed	Epoch AI	Claude 3.5 Sonnet (October 2024)
Claude 3.5 Sonnet	54%	Anthropic	closed	Epoch AI	Claude 3.5 Sonnet
Grok-2	53.8%	xAI	closed	Epoch AI	Grok-2
Qwen3-4B	52.3%	Alibaba	open	Epoch AI	Qwen3-4B
Llama 4 Scout	51.8%	Meta	open	Epoch AI	Llama 4 Scout
Mistral Large 2	51.3%	Mistral AI	open	Epoch AI	Mistral Large 2
Llama 3.1-405B	50.9%	Meta	open	Epoch AI	Llama 3.1-405B
o1-preview	50.3%	OpenAI	closed	Epoch AI	o1-preview
GPT-4o	49.2%	OpenAI	closed	Epoch AI	GPT-4o
Mistral Small 3.2	49.1%	Mistral AI	open	Epoch AI	Mistral Small 3.2
Qwen2.5-72B	49.1%	Alibaba	open	Epoch AI	Qwen2.5-72B
GPT-4.1 Nano	48.9%	OpenAI	closed	Epoch AI	GPT-4.1 Nano
Qwen-Plus	48.1%	Alibaba	closed	Epoch AI	Qwen-Plus
Gemma 3 27B	47.7%	Google	open	Epoch AI	Gemma 3 27B
Magistral Small 1.2	47.6%	Mistral AI	open	Epoch AI	Magistral Small 1.2
Mistral Small 3.1	47.5%	Mistral AI	open	Epoch AI	Mistral Small 3.1
Llama 3.3 70B	47.4%	Meta	open	Epoch AI	Llama 3.3 70B
Mistral Small 3	47.3%	Mistral AI	open	Epoch AI	Mistral Small 3
Gemini 1.5 Flash	47.3%	Google	closed	Epoch AI	Gemini 1.5 Flash
Claude 3 Opus	47.2%	Anthropic	closed	Epoch AI	Claude 3 Opus
GPT-4 Turbo (Apr 2024)	46.6%	OpenAI	closed	Epoch AI	GPT-4 Turbo (Apr 2024)
Tulu 3 (Tülu 3) 70B	46.3%	Ai2 (Allen Institute)	open	Epoch AI	Tulu 3 (Tülu 3) 70B
Qwen2.5-32B	46.1%	Alibaba	open	Epoch AI	Qwen2.5-32B
DeepSeek-R1-Distill-Qwen-14B	44.7%	DeepSeek	open	Epoch AI	DeepSeek-R1-Distill-Qwen-14B
Llama 3.1-70B	44.2%	Meta	open	Epoch AI	Llama 3.1-70B
WizardLM-2 8x22B	43.4%	Microsoft	open	Epoch AI	WizardLM-2 8x22B
GPT-4 Turbo (Nov 2023)	42.4%	OpenAI	closed	Epoch AI	GPT-4 Turbo (Nov 2023)
Qwen-Turbo	41.8%	Alibaba	closed	Epoch AI	Qwen-Turbo
Llama 3.2 90B	41%	Meta	open	Epoch AI	Llama 3.2 90B
Qwen2-72B	40.8%	Alibaba	open	Epoch AI	Qwen2-72B
Claude 3 Sonnet	40.6%	Anthropic	closed	Epoch AI	Claude 3 Sonnet
Llama 3-70B	40.6%	Meta	open	Epoch AI	Llama 3-70B
Gemma 3 12B	39.5%	Google	open	Epoch AI	Gemma 3 12B
Mistral Large	38.8%	Mistral AI	closed	Epoch AI	Mistral Large
Claude 3.5 Haiku	38.1%	Anthropic	closed	Epoch AI	Claude 3.5 Haiku
Qwen3-1.7B	38%	Alibaba	open	Epoch AI	Qwen3-1.7B
GPT-4o-mini	37.7%	OpenAI	closed	Epoch AI	GPT-4o-mini
Gemma 2 27B	36.5%	Google	open	Epoch AI	Gemma 2 27B
Claude 3 Haiku	36.3%	Anthropic	closed	Epoch AI	Claude 3 Haiku
GPT-4	35.7%	OpenAI	closed	Epoch AI	GPT-4
Qwen2.5-7B	35.5%	Alibaba	open	Epoch AI	Qwen2.5-7B
Claude 2	34.7%	Anthropic	closed	Epoch AI	Claude 2
Mixtral 8x22B	34.1%	Mistral AI	open	Epoch AI	Mixtral 8x22B
Gemini 1.0 Pro	34%	Google	closed	Epoch AI	Gemini 1.0 Pro
DeepSeek-R1-Distill-Qwen-1.5B	33.6%	DeepSeek	open	Epoch AI	DeepSeek-R1-Distill-Qwen-1.5B
Claude 2.1	33%	Anthropic	closed	Epoch AI	Claude 2.1
Gemini 1.5 Flash 8B	33%	Google	closed	Epoch AI	Gemini 1.5 Flash 8B
Yi-1.5-34B	32%	01.AI	open	Epoch AI	Yi-1.5-34B
Qwen1.5-32B	30.7%	Alibaba	open	Epoch AI	Qwen1.5-32B
Mixtral 8x7B	30.6%	Mistral AI	open	Epoch AI	Mixtral 8x7B
Mistral Nemo	29.9%	Mistral AI	open	Epoch AI	Mistral Nemo
Qwen1.5-72B	28.8%	Alibaba	open	Epoch AI	Qwen1.5-72B
GPT-3.5 Turbo	28%	OpenAI	closed	Epoch AI	GPT-3.5 Turbo
phi-3-medium 14B	27.6%	Microsoft	open	Epoch AI	phi-3-medium 14B
Gemma 2 9B	27.5%	Google	open	Epoch AI	Gemma 2 9B
Ministral 8B	27.1%	Mistral AI	open	Epoch AI	Ministral 8B
Llama 3.1-8B	27%	Meta	open	Epoch AI	Llama 3.1-8B
Llama 2-70B	26.3%	Meta	open	Epoch AI	Llama 2-70B
Llama 3-8B	26.1%	Meta	open	Epoch AI	Llama 3-8B
Ministral 3B	25.3%	Mistral AI	closed	Epoch AI	Ministral 3B
DeepSeek LLM 67B	24.6%	DeepSeek	open	Epoch AI	DeepSeek LLM 67B
Llama 3.2 1B	23.9%	Meta	open	Epoch AI	Llama 3.2 1B
Gemma 3 4B	23.2%	Google	open	Epoch AI	Gemma 3 4B
Gemma 3 1B	19.9%	Google	open	Epoch AI	Gemma 3 1B
Mistral 7B	15.2%	Mistral AI	open	Epoch AI	Mistral 7B
Yi-34B	14.7%	01.AI	open	Epoch AI	Yi-34B

OTIS Mock AIME

How to read each metric

Search modelLabWeights

OTIS Mock AIME
Model	%	Lab	Weights	Source	Compare models
Claude Opus 5.5	100%	Anthropic	closed	Epoch AI	Claude Opus 5.5
Claude Sonnet 5.5	100%	Anthropic	closed	Epoch AI	Claude Sonnet 5.5
GPT-6.1 Sol	100%	OpenAI	closed	Epoch AI	GPT-6.1 Sol
GPT-6 Astra	100%	OpenAI	closed	Epoch AI	GPT-6 Astra
GPT-5.5	100%	OpenAI	closed	Epoch AI	GPT-5.5
Gemini 4 Argon	100%	Google	closed	Epoch AI	Gemini 4 Argon
Qwen3.8 Max	100%	Alibaba	closed	Epoch AI	Qwen3.8 Max
Kimi K3.1 Pro	100%	Moonshot AI	open	Epoch AI	Kimi K3.1 Pro
Claude Fable 5	100%	Anthropic	closed	Epoch AI	Claude Fable 5
Claude Fable 5.1	100%	Anthropic	closed	Epoch AI	Claude Fable 5.1
GPT-5.5 Pro	100%	OpenAI	closed	Epoch AI	GPT-5.5 Pro
GPT-5.6 Sol	100%	OpenAI	closed	Epoch AI	GPT-5.6 Sol
GPT-6 Sol	100%	OpenAI	closed	Epoch AI	GPT-6 Sol
GPT-6.1 Terra	100%	OpenAI	closed	Epoch AI	GPT-6.1 Terra
Claude Fable 5.2	99.7%	Anthropic	closed	Epoch AI	Claude Fable 5.2
GPT-5.6 Terra	99.7%	OpenAI	closed	Epoch AI	GPT-5.6 Terra
DeepSeek V4.1 Flash	99.4%	DeepSeek	open	Epoch AI	DeepSeek V4.1 Flash
Mistral Large 4	99.4%	Mistral AI	open	Epoch AI	Mistral Large 4
Muse Wonder	99.4%	Meta	closed	Epoch AI	Muse Wonder
Grok 4.6	99.2%	xAI	closed	Epoch AI	Grok 4.6
Muse Spark 1.3	99.2%	Meta	closed	Epoch AI	Muse Spark 1.3
Doubao Seed Evolving (Nov 2026)	99.2%	ByteDance	closed	Epoch AI	Doubao Seed Evolving (Nov 2026)
Claude Haiku 5.5	98.9%	Anthropic	closed	Epoch AI	Claude Haiku 5.5
GPT-6 Luna	98.9%	OpenAI	closed	Epoch AI	GPT-6 Luna
Gemini 3.8 Flash	98.9%	Google	closed	Epoch AI	Gemini 3.8 Flash
Claude Opus 5	98.9%	Anthropic	closed	Epoch AI	Claude Opus 5
DeepSeek V4 Pro	98.6%	DeepSeek	open	Epoch AI	DeepSeek V4 Pro
Claude Opus 4.8	98.3%	Anthropic	closed	Epoch AI	Claude Opus 4.8
GPT-5.6 Luna	98.3%	OpenAI	closed	Epoch AI	GPT-5.6 Luna
Grok 4.7	98.1%	xAI	closed	Epoch AI	Grok 4.7
Muse Spark Open	98.1%	Meta	open	Epoch AI	Muse Spark Open
Claude Opus 4.7	97.8%	Anthropic	closed	Epoch AI	Claude Opus 4.7
GPT-5.4	97.8%	OpenAI	closed	Epoch AI	GPT-5.4
Grok 4.5	97.8%	xAI	closed	Epoch AI	Grok 4.5
Kimi K3	97.2%	Moonshot AI	open	Epoch AI	Kimi K3
Gemini 3.7 Flash	97.2%	Google	closed	Epoch AI	Gemini 3.7 Flash
GPT-5.2	96.1%	OpenAI	closed	Epoch AI	GPT-5.2
Kimi K2.6	96.1%	Moonshot AI	open	Epoch AI	Kimi K2.6
Doubao Seed 2.1 Turbo	96.1%	ByteDance	closed	Epoch AI	Doubao Seed 2.1 Turbo
Gemini 3.1 Pro	95.6%	Google	closed	Epoch AI	Gemini 3.1 Pro
Gemini 3.5 Flash	95.6%	Google	closed	Epoch AI	Gemini 3.5 Flash
Kimi K2.7 Code	95.6%	Moonshot AI	open	Epoch AI	Kimi K2.7 Code
Qwen3.7 Max	95.6%	Alibaba	closed	Epoch AI	Qwen3.7 Max
Doubao Seed 2.1 Pro	95.6%	ByteDance	closed	Epoch AI	Doubao Seed 2.1 Pro
Gemini 3 Flash	95.6%	Google	closed	Epoch AI	Gemini 3 Flash
Claude Sonnet 5	94.7%	Anthropic	closed	Epoch AI	Claude Sonnet 5
Claude Opus 4.6	94.4%	Anthropic	closed	Epoch AI	Claude Opus 4.6
DeepSeek V4 Flash 0731	94.4%	DeepSeek	open	Epoch AI	DeepSeek V4 Flash 0731
Gemini 3.6 Flash	94.2%	Google	closed	Epoch AI	Gemini 3.6 Flash
GLM-5.3-Flash	93.9%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.3-Flash
Doubao-Seed-2.0 Pro	93.9%	ByteDance	closed	Epoch AI	Doubao-Seed-2.0 Pro
Qwen3.6 Plus	93.3%	Alibaba	closed	Epoch AI	Qwen3.6 Plus
Qwen3.7 Plus	93.3%	Alibaba	closed	Epoch AI	Qwen3.7 Plus
GLM 5.1	93.3%	Zhipu AI (Z.ai)	open	Epoch AI	GLM 5.1
Grok 4.3 Beta	93.3%	xAI	closed	Epoch AI	Grok 4.3 Beta
Kimi K2.5	92.2%	Moonshot AI	open	Epoch AI	Kimi K2.5
Grok 4.20	92.2%	xAI	closed	Epoch AI	Grok 4.20
GPT-5	91.4%	OpenAI	closed	Epoch AI	GPT-5
Gemini 3 Pro	91.4%	Google	closed	Epoch AI	Gemini 3 Pro
GLM-5.3	91.1%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.3
Qwen3.6 27B	91.1%	Alibaba	open	Epoch AI	Qwen3.6 27B
Doubao Seed 2.0 Lite (Apr 2026)	91.1%	ByteDance	closed	Epoch AI	Doubao Seed 2.0 Lite (Apr 2026)
Qwen 3.6 Max (Preview)	91.1%	Alibaba	undisclosed	Epoch AI	Qwen 3.6 Max (Preview)
gpt-oss-120b	88.9%	OpenAI	open	Epoch AI	gpt-oss-120b
GPT-5.4 Mini	88.9%	OpenAI	closed	Epoch AI	GPT-5.4 Mini
Qwen3.5 397B A17B	88.9%	Alibaba	open	Epoch AI	Qwen3.5 397B A17B
Muse Spark	88.9%	Meta	closed	Epoch AI	Muse Spark
GPT-5.1	88.6%	OpenAI	closed	Epoch AI	GPT-5.1
DeepSeek V3.2	87.8%	DeepSeek	open	Epoch AI	DeepSeek V3.2
GPT-5.4 Nano	87.8%	OpenAI	closed	Epoch AI	GPT-5.4 Nano
Qwen3-235B-A22B	86.7%	Alibaba	open	Epoch AI	Qwen3-235B-A22B
Nemotron 3 Ultra	86.7%	NVIDIA	open	Epoch AI	Nemotron 3 Ultra
GPT-5 Mini	86.7%	OpenAI	closed	Epoch AI	GPT-5 Mini
Qwen3.6 35B A3B	86.7%	Alibaba	open	Epoch AI	Qwen3.6 35B A3B
Qwen3.7 Flash	86.7%	Alibaba	closed	Epoch AI	Qwen3.7 Flash
Qwen 3.5 Plus (hosted 397B-A17B)	86.7%	Alibaba	undisclosed	Epoch AI	Qwen 3.5 Plus (hosted 397B-A17B)
GLM-5.2	86.4%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.2
Claude Opus 4.5	86.1%	Anthropic	closed	Epoch AI	Claude Opus 4.5
Claude Sonnet 4.6	85.8%	Anthropic	closed	Epoch AI	Claude Sonnet 4.6
Gemini 2.5 Pro	84.7%	Google	closed	Epoch AI	Gemini 2.5 Pro
o3	84.4%	OpenAI	closed	Epoch AI	o3
Qwen3.5-Flash	84.4%	Alibaba	undisclosed	Epoch AI	Qwen3.5-Flash
Qwen3.6 Flash	84.4%	Alibaba	undisclosed	Epoch AI	Qwen3.6 Flash
Grok 4	84%	xAI	closed	Epoch AI	Grok 4
GLM 4.7	83.3%	Zhipu AI (Z.ai)	open	Epoch AI	GLM 4.7
Kimi K2 Thinking	83.1%	Moonshot AI	open	Epoch AI	Kimi K2 Thinking
Gemma 4 26B A4B	82.2%	Google	open	Epoch AI	Gemma 4 26B A4B
o4-mini	81.7%	OpenAI	closed	Epoch AI	o4-mini
GPT-5 Nano	81.1%	OpenAI	closed	Epoch AI	GPT-5 Nano
Gemini 3.1 Flash Lite	80%	Google	closed	Epoch AI	Gemini 3.1 Flash Lite
GLM 5	80%	Zhipu AI (Z.ai)	open	Epoch AI	GLM 5
Claude Sonnet 4.5	77.8%	Anthropic	closed	Epoch AI	Claude Sonnet 4.5
Grok-3 mini	77.8%	xAI	closed	Epoch AI	Grok-3 mini
o3-mini	76.9%	OpenAI	closed	Epoch AI	o3-mini
Gemma 4 31B	73.3%	Google	open	Epoch AI	Gemma 4 31B
Qwen3-Max	73.3%	Alibaba	closed	Epoch AI	Qwen3-Max
o1	73.3%	OpenAI	closed	Epoch AI	o1
Gemini 2.5 Flash	73.1%	Google	closed	Epoch AI	Gemini 2.5 Flash
MiniMax M3	71.1%	MiniMax	open	Epoch AI	MiniMax M3
Gemini 3.5 Flash Lite	71.1%	Google	closed	Epoch AI	Gemini 3.5 Flash Lite
Claude Sonnet 4	71.1%	Anthropic	closed	Epoch AI	Claude Sonnet 4
Qwen3 30B A3B	70.3%	Alibaba	open	Epoch AI	Qwen3 30B A3B
Qwen3.5-35B-A3B	70%	Alibaba	open	Epoch AI	Qwen3.5-35B-A3B
Doubao-Seed-1.6	69.4%	ByteDance	closed	Epoch AI	Doubao-Seed-1.6
Claude Opus 4.1	68.9%	Anthropic	closed	Epoch AI	Claude Opus 4.1
GPT-5.5 Instant	68.1%	OpenAI	closed	Epoch AI	GPT-5.5 Instant
Qwen3 32B	66.9%	Alibaba	open	Epoch AI	Qwen3 32B
Claude Haiku 4.5	66.7%	Anthropic	closed	Epoch AI	Claude Haiku 4.5
DeepSeek R1	66.4%	DeepSeek	open	Epoch AI	DeepSeek R1
Qwen3 14B	66.4%	Alibaba	open	Epoch AI	Qwen3 14B
gpt-oss-20b	65.3%	OpenAI	open	Epoch AI	gpt-oss-20b
Claude Opus 4	64.4%	Anthropic	closed	Epoch AI	Claude Opus 4
Qwen3.5-9B	61.7%	Alibaba	open	Epoch AI	Qwen3.5-9B
QwQ-32B	59.2%	Alibaba	open	Epoch AI	QwQ-32B
Claude 3.7 Sonnet	57.8%	Anthropic	closed	Epoch AI	Claude 3.7 Sonnet
Gemini 2.0 Flash Thinking	57.8%	Google	closed	Epoch AI	Gemini 2.0 Flash Thinking
Qwen3-8B	56.1%	Alibaba	open	Epoch AI	Qwen3-8B
Qwen3.5-4B	55.8%	Alibaba	open	Epoch AI	Qwen3.5-4B
DeepSeek-R1-Distill-Qwen-32B	55.6%	DeepSeek	open	Epoch AI	DeepSeek-R1-Distill-Qwen-32B
Grok 3	55.6%	xAI	closed	Epoch AI	Grok 3
DeepSeek-R1-Distill-Llama-70B	51.4%	DeepSeek	open	Epoch AI	DeepSeek-R1-Distill-Llama-70B
DeepSeek-R1-Distill-Qwen-14B	50.6%	DeepSeek	open	Epoch AI	DeepSeek-R1-Distill-Qwen-14B
o1-mini	46.9%	OpenAI	closed	Epoch AI	o1-mini
GPT-4.1 Mini	44.7%	OpenAI	closed	Epoch AI	GPT-4.1 Mini
GPT-4.1	38.3%	OpenAI	closed	Epoch AI	GPT-4.1
DeepSeek V3	37.8%	DeepSeek	open	Epoch AI	DeepSeek V3
GPT-4.5	37.8%	OpenAI	closed	Epoch AI	GPT-4.5
Mistral Medium 3	32.2%	Mistral AI	closed	Epoch AI	Mistral Medium 3
Gemini 2.0 Flash	31.1%	Google	closed	Epoch AI	Gemini 2.0 Flash
o1-preview	31.1%	OpenAI	closed	Epoch AI	o1-preview
Mistral Small 3.2	30.3%	Mistral AI	open	Epoch AI	Mistral Small 3.2
Magistral Small 1.0	30%	Mistral AI	open	Epoch AI	Magistral Small 1.0
GPT-4.1 Nano	28.9%	OpenAI	closed	Epoch AI	GPT-4.1 Nano
Magistral Small 1.2	28.1%	Mistral AI	open	Epoch AI	Magistral Small 1.2
Gemini 1.5 Pro	23.1%	Google	closed	Epoch AI	Gemini 1.5 Pro
Gemma 3 27B	22.5%	Google	open	Epoch AI	Gemma 3 27B
DeepSeek-R1-Distill-Qwen-1.5B	21.4%	DeepSeek	open	Epoch AI	DeepSeek-R1-Distill-Qwen-1.5B
Llama 4 Maverick	20.6%	Meta	open	Epoch AI	Llama 4 Maverick
Qwen-Plus	17.8%	Alibaba	closed	Epoch AI	Qwen-Plus
Gemma 3 12B	16.7%	Google	open	Epoch AI	Gemma 3 12B
Gemini 1.5 Flash	16.2%	Google	closed	Epoch AI	Gemini 1.5 Flash
Qwen2.5-Max	16.1%	Alibaba	closed	Epoch AI	Qwen2.5-Max
DeepSeek V3	15.8%	DeepSeek	open	Epoch AI	DeepSeek V3
Phi 4	13.8%	Microsoft	open	Epoch AI	Phi 4
Grok-2	11.5%	xAI	closed	Epoch AI	Grok-2
Llama 3.1-405B	9.7%	Meta	open	Epoch AI	Llama 3.1-405B
Claude 3.5 Sonnet (October 2024)	8.5%	Anthropic	closed	Epoch AI	Claude 3.5 Sonnet (October 2024)
Mistral Large 2	8.5%	Mistral AI	open	Epoch AI	Mistral Large 2
Qwen2.5-72B	8.1%	Alibaba	open	Epoch AI	Qwen2.5-72B
Qwen3-1.7B	8.1%	Alibaba	open	Epoch AI	Qwen3-1.7B
Llama 4 Scout	7.8%	Meta	open	Epoch AI	Llama 4 Scout
Gemma 3 4B	7.5%	Google	open	Epoch AI	Gemma 3 4B
Qwen2.5-32B	7.4%	Alibaba	open	Epoch AI	Qwen2.5-32B
GPT-4o-mini	6.9%	OpenAI	closed	Epoch AI	GPT-4o-mini
Mistral Small 3	6.7%	Mistral AI	open	Epoch AI	Mistral Small 3
GPT-4 Turbo (Apr 2024)	6.7%	OpenAI	closed	Epoch AI	GPT-4 Turbo (Apr 2024)
Claude 3.5 Sonnet	6.5%	Anthropic	closed	Epoch AI	Claude 3.5 Sonnet
GPT-4o	6.4%	OpenAI	closed	Epoch AI	GPT-4o
Qwen-Turbo	6.1%	Alibaba	closed	Epoch AI	Qwen-Turbo
Mistral Small 3.1	5.8%	Mistral AI	open	Epoch AI	Mistral Small 3.1
Llama 3.3 70B	5.1%	Meta	open	Epoch AI	Llama 3.3 70B
Claude 3 Opus	4.7%	Anthropic	closed	Epoch AI	Claude 3 Opus
Gemini 1.5 Flash 8B	4.6%	Google	closed	Epoch AI	Gemini 1.5 Flash 8B
Tulu 3 (Tülu 3) 70B	4.4%	Ai2 (Allen Institute)	open	Epoch AI	Tulu 3 (Tülu 3) 70B
Claude 3.5 Haiku	4.3%	Anthropic	closed	Epoch AI	Claude 3.5 Haiku
Llama 3-70B	4.3%	Meta	open	Epoch AI	Llama 3-70B
Llama 3.1-70B	3.6%	Meta	open	Epoch AI	Llama 3.1-70B
Llama 3.2 90B	2.6%	Meta	open	Epoch AI	Llama 3.2 90B
Claude 2	2.5%	Anthropic	closed	Epoch AI	Claude 2
Claude 3 Sonnet	2.5%	Anthropic	closed	Epoch AI	Claude 3 Sonnet
Qwen2.5-7B	2.5%	Alibaba	open	Epoch AI	Qwen2.5-7B
GPT-3.5 Turbo	2.2%	OpenAI	closed	Epoch AI	GPT-3.5 Turbo
Mistral Large	1.9%	Mistral AI	closed	Epoch AI	Mistral Large
Claude 2.1	1.9%	Anthropic	closed	Epoch AI	Claude 2.1
Llama 3-8B	1.9%	Meta	open	Epoch AI	Llama 3-8B
Claude 3 Haiku	1.8%	Anthropic	closed	Epoch AI	Claude 3 Haiku
Llama 3.1-8B	1.7%	Meta	open	Epoch AI	Llama 3.1-8B
Gemma 2 27B	1.4%	Google	open	Epoch AI	Gemma 2 27B
GPT-4	1.1%	OpenAI	closed	Epoch AI	GPT-4
Gemini 1.0 Pro	1.1%	Google	closed	Epoch AI	Gemini 1.0 Pro
Gemma 3 1B	1.1%	Google	open	Epoch AI	Gemma 3 1B
DeepSeek LLM 67B	0.8%	DeepSeek	open	Epoch AI	DeepSeek LLM 67B
Gemma 2 9B	0.6%	Google	open	Epoch AI	Gemma 2 9B
Llama 3.2 1B	0.6%	Meta	open	Epoch AI	Llama 3.2 1B
Mistral 7B	0.3%	Mistral AI	open	Epoch AI	Mistral 7B
Llama 2-70B	0%	Meta	open	Epoch AI	Llama 2-70B

FrontierMath (tiers 1 to 3)

How to read each metric

Search modelLabWeights

FrontierMath (tiers 1 to 3)
Model	%	Lab	Weights	Source	Compare models
Claude Fable 5.2	95.1%	Anthropic	closed	Epoch AI	Claude Fable 5.2
GPT-6.1 Sol	93.7%	OpenAI	closed	Epoch AI	GPT-6.1 Sol
GPT-6 Astra	93.7%	OpenAI	closed	Epoch AI	GPT-6 Astra
Claude Opus 5.5	91.2%	Anthropic	closed	Epoch AI	Claude Opus 5.5
GPT-6.1 Terra	90.9%	OpenAI	closed	Epoch AI	GPT-6.1 Terra
Claude Fable 5.1	90.2%	Anthropic	closed	Epoch AI	Claude Fable 5.1
GPT-6 Sol	89.8%	OpenAI	closed	Epoch AI	GPT-6 Sol
GPT-5.6 Sol	89.1%	OpenAI	closed	Epoch AI	GPT-5.6 Sol
Claude Sonnet 5.5	88.8%	Anthropic	closed	Epoch AI	Claude Sonnet 5.5
GPT-5.5 Pro	87.7%	OpenAI	closed	Epoch AI	GPT-5.5 Pro
Claude Fable 5	87%	Anthropic	closed	Epoch AI	Claude Fable 5
Gemini 4 Argon	86%	Google	closed	Epoch AI	Gemini 4 Argon
GPT-5.6 Terra	86%	OpenAI	closed	Epoch AI	GPT-5.6 Terra
Claude Opus 5	85.6%	Anthropic	closed	Epoch AI	Claude Opus 5
GPT-5.5	85.3%	OpenAI	closed	Epoch AI	GPT-5.5
GPT-5.4 Pro	82.5%	OpenAI	closed	Epoch AI	GPT-5.4 Pro
Muse Wonder	82.1%	Meta	closed	Epoch AI	Muse Wonder
GPT-5.6 Luna	82.1%	OpenAI	closed	Epoch AI	GPT-5.6 Luna
Claude Opus 4.8	80%	Anthropic	closed	Epoch AI	Claude Opus 4.8
GPT-6 Luna	78.9%	OpenAI	closed	Epoch AI	GPT-6 Luna
GPT-5.4	78.6%	OpenAI	closed	Epoch AI	GPT-5.4
Kimi K3.1 Pro	78.2%	Moonshot AI	open	Epoch AI	Kimi K3.1 Pro
Claude Haiku 5.5	75.1%	Anthropic	closed	Epoch AI	Claude Haiku 5.5
Qwen3.8 Max	74.7%	Alibaba	closed	Epoch AI	Qwen3.8 Max
Muse Spark 1.3	74.4%	Meta	closed	Epoch AI	Muse Spark 1.3
GPT-5.2 Pro	74%	OpenAI	closed	Epoch AI	GPT-5.2 Pro
Kimi K3	72.2%	Moonshot AI	open	Epoch AI	Kimi K3
Gemini 3.7 Flash	71.6%	Google	closed	Epoch AI	Gemini 3.7 Flash
DeepSeek V4.1 Flash	71.2%	DeepSeek	open	Epoch AI	DeepSeek V4.1 Flash
Claude Opus 4.7	70.2%	Anthropic	closed	Epoch AI	Claude Opus 4.7
Muse Spark Open	69.8%	Meta	open	Epoch AI	Muse Spark Open
GLM-5.3	68.8%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.3
Gemini 3.8 Flash	68.4%	Google	closed	Epoch AI	Gemini 3.8 Flash
GPT-5.2	67.4%	OpenAI	closed	Epoch AI	GPT-5.2
Grok 4.6	66%	xAI	closed	Epoch AI	Grok 4.6
Claude Opus 4.6	66%	Anthropic	closed	Epoch AI	Claude Opus 4.6
Claude Sonnet 5	65.6%	Anthropic	closed	Epoch AI	Claude Sonnet 5
DeepSeek V4 Pro	64.6%	DeepSeek	open	Epoch AI	DeepSeek V4 Pro
Qwen3.7 Max	64.6%	Alibaba	closed	Epoch AI	Qwen3.7 Max
Gemini 3.5 Flash	62.8%	Google	closed	Epoch AI	Gemini 3.5 Flash
Gemini 3.1 Pro	59.6%	Google	closed	Epoch AI	Gemini 3.1 Pro
GLM-5.2	59.2%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.2
Gemini 3.6 Flash	58.9%	Google	closed	Epoch AI	Gemini 3.6 Flash
Mistral Large 4	58.6%	Mistral AI	open	Epoch AI	Mistral Large 4
DeepSeek V4 Flash 0731	57.5%	DeepSeek	open	Epoch AI	DeepSeek V4 Flash 0731
Kimi K2.6	57.2%	Moonshot AI	open	Epoch AI	Kimi K2.6
Grok 4.5	57.2%	xAI	closed	Epoch AI	Grok 4.5
GLM-5.3-Flash	55.8%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.3-Flash
GPT-5 Pro	55.8%	OpenAI	closed	Epoch AI	GPT-5 Pro
GPT-5	55.4%	OpenAI	closed	Epoch AI	GPT-5
Kimi K2.7 Code	54%	Moonshot AI	open	Epoch AI	Kimi K2.7 Code
Grok 4.7	53%	xAI	closed	Epoch AI	Grok 4.7
GPT-5.4 Mini	51.2%	OpenAI	closed	Epoch AI	GPT-5.4 Mini
Gemini 3 Flash	51.2%	Google	closed	Epoch AI	Gemini 3 Flash
GPT-5 Mini	46.7%	OpenAI	closed	Epoch AI	GPT-5 Mini
GPT-5.4 Nano	44.9%	OpenAI	closed	Epoch AI	GPT-5.4 Nano
Grok 4.20	44.9%	xAI	closed	Epoch AI	Grok 4.20
Grok 4.3 Beta	42.8%	xAI	closed	Epoch AI	Grok 4.3 Beta
Qwen3.6 Plus	38.2%	Alibaba	closed	Epoch AI	Qwen3.6 Plus
GLM 5.1	36.8%	Zhipu AI (Z.ai)	open	Epoch AI	GLM 5.1
o4-mini	36.1%	OpenAI	closed	Epoch AI	o4-mini
Qwen3.6 27B	35.1%	Alibaba	open	Epoch AI	Qwen3.6 27B
Claude Opus 4.5	34.4%	Anthropic	closed	Epoch AI	Claude Opus 4.5
Qwen3.7 Plus	34.4%	Alibaba	closed	Epoch AI	Qwen3.7 Plus
o3	33.3%	OpenAI	closed	Epoch AI	o3
Qwen3.5 397B A17B	31.2%	Alibaba	open	Epoch AI	Qwen3.5 397B A17B
Gemini 3.1 Flash Lite	27.7%	Google	closed	Epoch AI	Gemini 3.1 Flash Lite
GPT-5.5 Instant	26.3%	OpenAI	closed	Epoch AI	GPT-5.5 Instant
Gemini 3.5 Flash Lite	26%	Google	closed	Epoch AI	Gemini 3.5 Flash Lite
Gemini 2.5 Pro	24.6%	Google	closed	Epoch AI	Gemini 2.5 Pro
Claude Sonnet 4.5	23.9%	Anthropic	closed	Epoch AI	Claude Sonnet 4.5
Qwen3.6 Flash	22.5%	Alibaba	undisclosed	Epoch AI	Qwen3.6 Flash
Qwen3.6 35B A3B	20.4%	Alibaba	open	Epoch AI	Qwen3.6 35B A3B
GPT-5 Nano	20%	OpenAI	closed	Epoch AI	GPT-5 Nano
Qwen3.7 Flash	19.3%	Alibaba	closed	Epoch AI	Qwen3.7 Flash
Qwen3-Max	18.9%	Alibaba	closed	Epoch AI	Qwen3-Max
o3-mini	18.6%	OpenAI	closed	Epoch AI	o3-mini
Qwen3.5-Flash	18.2%	Alibaba	undisclosed	Epoch AI	Qwen3.5-Flash
o1	14.7%	OpenAI	closed	Epoch AI	o1
Claude Opus 4.1	12.6%	Anthropic	closed	Epoch AI	Claude Opus 4.1
GPT-4.1 Mini	6.7%	OpenAI	closed	Epoch AI	GPT-4.1 Mini
GPT-4.1	6%	OpenAI	closed	Epoch AI	GPT-4.1
GPT-4o-mini	0.7%	OpenAI	closed	Epoch AI	GPT-4o-mini
GPT-4 Turbo (Apr 2024)	0.7%	OpenAI	closed	Epoch AI	GPT-4 Turbo (Apr 2024)
GPT-4o	0.4%	OpenAI	closed	Epoch AI	GPT-4o
GPT-3.5 Turbo	0%	OpenAI	closed	Epoch AI	GPT-3.5 Turbo

SimpleQA Verified

How to read each metric

Search modelLabWeights

SimpleQA Verified
Model	%	Lab	Weights	Source	Compare models
Claude Fable 5.2	94%	Anthropic	closed	Epoch AI	Claude Fable 5.2
Gemini 4 Argon	76%	Google	closed	Epoch AI	Gemini 4 Argon
GPT-6 Astra	75.6%	OpenAI	closed	Epoch AI	GPT-6 Astra
GPT-6.1 Sol	73.9%	OpenAI	closed	Epoch AI	GPT-6.1 Sol
Gemini 3.1 Pro	73.5%	Google	closed	Epoch AI	Gemini 3.1 Pro
Claude Opus 5.5	72.2%	Anthropic	closed	Epoch AI	Claude Opus 5.5
Claude Fable 5.1	70.8%	Anthropic	closed	Epoch AI	Claude Fable 5.1
Claude Fable 5	70.7%	Anthropic	closed	Epoch AI	Claude Fable 5
Gemini 3.8 Flash	69.7%	Google	closed	Epoch AI	Gemini 3.8 Flash
GPT-5.6 Sol	69.7%	OpenAI	closed	Epoch AI	GPT-5.6 Sol
Gemini 3.7 Flash	69.2%	Google	closed	Epoch AI	Gemini 3.7 Flash
Gemini 3 Flash	66.8%	Google	closed	Epoch AI	Gemini 3 Flash
Gemini 3.5 Flash	66.2%	Google	closed	Epoch AI	Gemini 3.5 Flash
Gemini 3.6 Flash	66.2%	Google	closed	Epoch AI	Gemini 3.6 Flash
GPT-5.5	63%	OpenAI	closed	Epoch AI	GPT-5.5
Kimi K3.1 Pro	62.2%	Moonshot AI	open	Epoch AI	Kimi K3.1 Pro
GPT-6 Sol	60.7%	OpenAI	closed	Epoch AI	GPT-6 Sol
Muse Spark 1.2	60.3%	Meta	closed	Epoch AI	Muse Spark 1.2
Claude Opus 5	59.9%	Anthropic	closed	Epoch AI	Claude Opus 5
Muse Spark 1.1	57.8%	Meta	closed	Epoch AI	Muse Spark 1.1
GPT-6.1 Terra	57.3%	OpenAI	closed	Epoch AI	GPT-6.1 Terra
Muse Spark Open	56.6%	Meta	open	Epoch AI	Muse Spark Open
Grok 4.7	56%	xAI	closed	Epoch AI	Grok 4.7
Qwen3.7 Max	55.8%	Alibaba	closed	Epoch AI	Qwen3.7 Max
Claude Opus 4.8	53%	Anthropic	closed	Epoch AI	Claude Opus 4.8
DeepSeek V4 Pro	52.9%	DeepSeek	open	Epoch AI	DeepSeek V4 Pro
Qwen 3.6 Max (Preview)	52%	Alibaba	undisclosed	Epoch AI	Qwen 3.6 Max (Preview)
Claude Opus 4.7	51.7%	Anthropic	closed	Epoch AI	Claude Opus 4.7
Kimi K3	50.6%	Moonshot AI	open	Epoch AI	Kimi K3
GPT-5	50.1%	OpenAI	closed	Epoch AI	GPT-5
o3	49.4%	OpenAI	closed	Epoch AI	o3
Grok 4.6	49.3%	xAI	closed	Epoch AI	Grok 4.6
Qwen3-Max	48.7%	Alibaba	closed	Epoch AI	Qwen3-Max
Grok 4.5	48.3%	xAI	closed	Epoch AI	Grok 4.5
GPT-5.1	48%	OpenAI	closed	Epoch AI	GPT-5.1
Qwen3.8 Max	47.3%	Alibaba	closed	Epoch AI	Qwen3.8 Max
Claude Opus 4.6	47%	Anthropic	closed	Epoch AI	Claude Opus 4.6
Claude Sonnet 5.5	46.5%	Anthropic	closed	Epoch AI	Claude Sonnet 5.5
GPT-5.4 Pro	46.3%	OpenAI	closed	Epoch AI	GPT-5.4 Pro
Claude Opus 4.5	45.7%	Anthropic	closed	Epoch AI	Claude Opus 4.5
GPT-5.4	45.1%	OpenAI	closed	Epoch AI	GPT-5.4
Doubao Seed 2.1 Pro	44.7%	ByteDance	closed	Epoch AI	Doubao Seed 2.1 Pro
Qwen3.6 Plus	44.1%	Alibaba	closed	Epoch AI	Qwen3.6 Plus
GPT-5.6 Terra	43.2%	OpenAI	closed	Epoch AI	GPT-5.6 Terra
DeepSeek V4.1 Flash	42.9%	DeepSeek	open	Epoch AI	DeepSeek V4.1 Flash
Doubao Seed Evolving (Nov 2026)	41.9%	ByteDance	closed	Epoch AI	Doubao Seed Evolving (Nov 2026)
GPT-6 Luna	41.4%	OpenAI	closed	Epoch AI	GPT-6 Luna
o1	41.1%	OpenAI	closed	Epoch AI	o1
GLM-5.3	41%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.3
GPT-5.6 Luna	41%	OpenAI	closed	Epoch AI	GPT-5.6 Luna
Qwen3-235B-A22B	40.4%	Alibaba	open	Epoch AI	Qwen3-235B-A22B
Doubao-Seed-2.0 Pro	39.4%	ByteDance	closed	Epoch AI	Doubao-Seed-2.0 Pro
GPT-5.2	37.1%	OpenAI	closed	Epoch AI	GPT-5.2
Kimi K2.7 Code	36.5%	Moonshot AI	open	Epoch AI	Kimi K2.7 Code
Claude Sonnet 4.6	35.5%	Anthropic	closed	Epoch AI	Claude Sonnet 4.6
Kimi K2.6	34.9%	Moonshot AI	open	Epoch AI	Kimi K2.6
Kimi K2.5	34.3%	Moonshot AI	open	Epoch AI	Kimi K2.5
GLM-5.2	34.2%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.2
GLM 5.1	34%	Zhipu AI (Z.ai)	open	Epoch AI	GLM 5.1
Claude Sonnet 5	33.7%	Anthropic	closed	Epoch AI	Claude Sonnet 5
DeepSeek V4 Flash 0731	33.6%	DeepSeek	open	Epoch AI	DeepSeek V4 Flash 0731
Grok 4.3 Beta	33.2%	xAI	closed	Epoch AI	Grok 4.3 Beta
GLM 4.7	32.2%	Zhipu AI (Z.ai)	open	Epoch AI	GLM 4.7
Doubao Seed 2.0 Lite (Apr 2026)	31.5%	ByteDance	closed	Epoch AI	Doubao Seed 2.0 Lite (Apr 2026)
GPT-4.1	31.1%	OpenAI	closed	Epoch AI	GPT-4.1
Claude Sonnet 4.5	30.7%	Anthropic	closed	Epoch AI	Claude Sonnet 4.5
Grok 4.20	30.2%	xAI	closed	Epoch AI	Grok 4.20
Doubao Seed 2.1 Turbo	29.7%	ByteDance	closed	Epoch AI	Doubao Seed 2.1 Turbo
GPT-5.4 Mini	29.4%	OpenAI	closed	Epoch AI	GPT-5.4 Mini
GPT-4o	26%	OpenAI	closed	Epoch AI	GPT-4o
Qwen 3.5 Plus (hosted 397B-A17B)	25.4%	Alibaba	undisclosed	Epoch AI	Qwen 3.5 Plus (hosted 397B-A17B)
Claude Haiku 5.5	23.8%	Anthropic	closed	Epoch AI	Claude Haiku 5.5
Mistral Large 4	22.4%	Mistral AI	open	Epoch AI	Mistral Large 4
GPT-5 Mini	21.6%	OpenAI	closed	Epoch AI	GPT-5 Mini
Qwen3.5-Flash	20.3%	Alibaba	undisclosed	Epoch AI	Qwen3.5-Flash
o4-mini	19.6%	OpenAI	closed	Epoch AI	o4-mini
Doubao-Seed-1.6	17.4%	ByteDance	closed	Epoch AI	Doubao-Seed-1.6
Qwen3.6 Flash	15.9%	Alibaba	undisclosed	Epoch AI	Qwen3.6 Flash
o3-mini	15.3%	OpenAI	closed	Epoch AI	o3-mini
Claude Haiku 4.5	13.2%	Anthropic	closed	Epoch AI	Claude Haiku 4.5
GPT-4.1 Mini	12.7%	OpenAI	closed	Epoch AI	GPT-4.1 Mini
Claude 3 Opus	12.6%	Anthropic	closed	Epoch AI	Claude 3 Opus
GPT-5 Nano	11.7%	OpenAI	closed	Epoch AI	GPT-5 Nano
GPT-5.4 Nano	11.7%	OpenAI	closed	Epoch AI	GPT-5.4 Nano
Gemma 4 31B	10.4%	Google	open	Epoch AI	Gemma 4 31B
GPT-4o-mini	8.3%	OpenAI	closed	Epoch AI	GPT-4o-mini
GPT-4.1 Nano	6%	OpenAI	closed	Epoch AI	GPT-4.1 Nano

SWE-bench Verified

How to read each metric

Search modelLabWeights

SWE-bench Verified
Model	%	Lab	Weights	Source	Compare models
Claude Opus 4.7	83.5%	Anthropic	closed	Epoch AI	Claude Opus 4.7
GPT-5.5	80.6%	OpenAI	closed	Epoch AI	GPT-5.5
Gemini 3.5 Flash	79.3%	Google	closed	Epoch AI	Gemini 3.5 Flash
GLM-5.2	78.7%	Zhipu AI (Z.ai)	open	Epoch AI	GLM-5.2
Claude Opus 4.6	78.7%	Anthropic	closed	Epoch AI	Claude Opus 4.6
DeepSeek V4 Pro	77.6%	DeepSeek	open	Epoch AI	DeepSeek V4 Pro
Qwen3.7 Max	77.3%	Alibaba	closed	Epoch AI	Qwen3.7 Max
GPT-5.4	76.9%	OpenAI	closed	Epoch AI	GPT-5.4
Kimi K2.6	76.7%	Moonshot AI	open	Epoch AI	Kimi K2.6
Claude Opus 4.5	76.7%	Anthropic	closed	Epoch AI	Claude Opus 4.5
Qwen 3.6 Max (Preview)	76.7%	Alibaba	undisclosed	Epoch AI	Qwen 3.6 Max (Preview)
Gemini 3.1 Pro	75.6%	Google	closed	Epoch AI	Gemini 3.1 Pro
Gemini 3 Flash	75.4%	Google	closed	Epoch AI	Gemini 3 Flash
Claude Sonnet 4.6	75.2%	Anthropic	closed	Epoch AI	Claude Sonnet 4.6
GPT-5.3-Codex	74.8%	OpenAI	closed	Epoch AI	GPT-5.3-Codex
GLM 5.1	74.2%	Zhipu AI (Z.ai)	open	Epoch AI	GLM 5.1
GPT-5.2	73.8%	OpenAI	closed	Epoch AI	GPT-5.2
Kimi K2.5	73.8%	Moonshot AI	open	Epoch AI	Kimi K2.5
GPT-5	73.6%	OpenAI	closed	Epoch AI	GPT-5
Claude Opus 4.1	73.3%	Anthropic	closed	Epoch AI	Claude Opus 4.1
Gemini 3 Pro	72.9%	Google	closed	Epoch AI	Gemini 3 Pro
GLM 5	72.1%	Zhipu AI (Z.ai)	open	Epoch AI	GLM 5
Claude Sonnet 4.5	71.3%	Anthropic	closed	Epoch AI	Claude Sonnet 4.5
Claude Opus 4	70.7%	Anthropic	closed	Epoch AI	Claude Opus 4
GPT-5.1	68%	OpenAI	closed	Epoch AI	GPT-5.1
GPT-5 Mini	64.7%	OpenAI	closed	Epoch AI	GPT-5 Mini
o3	62.3%	OpenAI	closed	Epoch AI	o3
Claude 3.7 Sonnet	61%	Anthropic	closed	Epoch AI	Claude 3.7 Sonnet
Qwen3.6 Plus	57.9%	Alibaba	closed	Epoch AI	Qwen3.6 Plus
Gemini 2.5 Pro	57.6%	Google	closed	Epoch AI	Gemini 2.5 Pro
GPT-4.1	48.5%	OpenAI	closed	Epoch AI	GPT-4.1
GPT-4o	31%	OpenAI	closed	Epoch AI	GPT-4o

GPQA Diamond (OpenRouter)

How to read each metric

Search modelLabWeights

GPQA Diamond (OpenRouter)
Model	%	Lab	Weights	Source	Compare models
Gemini 4 Argon	96.5%	Google	closed	OpenRouter Benchmarks	Gemini 4 Argon
Gemini 3.8 Flash	95.6%	Google	closed	OpenRouter Benchmarks	Gemini 3.8 Flash
GPT-6 Astra Pro	95.5%	OpenAI	undisclosed	OpenRouter Benchmarks	GPT-6 Astra Pro
Gemini 3.1 Pro	95.2%	Google	closed	OpenRouter Benchmarks	Gemini 3.1 Pro
GPT-6.1 Sol	94.8%	OpenAI	closed	OpenRouter Benchmarks	GPT-6.1 Sol
Ling 3.1 Flash	94.8%	Ant Group	undisclosed	OpenRouter Benchmarks	Ling 3.1 Flash
GPT-6 Astra	94.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-6 Astra
Gemini 3.7 Flash	94.3%	Google	closed	OpenRouter Benchmarks	Gemini 3.7 Flash
GPT-5.6 Sol Pro	94.1%	OpenAI	undisclosed	OpenRouter Benchmarks	GPT-5.6 Sol Pro
GPT-5.5	93.9%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.5
Kimi K3.1 Pro	93.9%	Moonshot AI	open	OpenRouter Benchmarks	Kimi K3.1 Pro
Kimi K3	93.5%	Moonshot AI	open	OpenRouter Benchmarks	Kimi K3
Gemini 3.6 Flash	93.4%	Google	closed	OpenRouter Benchmarks	Gemini 3.6 Flash
Claude Sonnet 5.5	93%	Anthropic	closed	OpenRouter Benchmarks	Claude Sonnet 5.5
DeepSeek V4.1 Flash	92.8%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V4.1 Flash
Grok 4.6	92.8%	xAI	closed	OpenRouter Benchmarks	Grok 4.6
GPT-5.6 Sol	92.8%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.6 Sol
Gemini 3.5 Flash	92.6%	Google	closed	OpenRouter Benchmarks	Gemini 3.5 Flash
GLM-5.3-Flash	92.5%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM-5.3-Flash
GPT-6 Sol	92.5%	OpenAI	closed	OpenRouter Benchmarks	GPT-6 Sol
Nemotron 3.5 Super VL	92.2%	NVIDIA	open	OpenRouter Benchmarks	Nemotron 3.5 Super VL
DeepSeek V4 Pro	91.9%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V4 Pro
Grok 4.7	91.9%	xAI	closed	OpenRouter Benchmarks	Grok 4.7
GPT-6.1 Terra	91.9%	OpenAI	closed	OpenRouter Benchmarks	GPT-6.1 Terra
Qwen3.8 2.4T A95B	91.8%	Alibaba	open	OpenRouter Benchmarks	Qwen3.8 2.4T A95B
MiniMax M3	91.3%	MiniMax	open	OpenRouter Benchmarks	MiniMax M3
Muse Spark Open	91%	Meta	open	OpenRouter Benchmarks	Muse Spark Open
GPT-5.6 Luna Pro	91%	OpenAI	undisclosed	OpenRouter Benchmarks	GPT-5.6 Luna Pro
DeepSeek V4 Flash Vision Exp	90.9%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V4 Flash Vision Exp
Kimi K2.7 Code	90.9%	Moonshot AI	open	OpenRouter Benchmarks	Kimi K2.7 Code
Qwen3.7 Max	90.9%	Alibaba	closed	OpenRouter Benchmarks	Qwen3.7 Max
Hy4	90.7%	Tencent	open	OpenRouter Benchmarks	Hy4
MiMo V2.6 Pro	90.6%	Xiaomi	open	OpenRouter Benchmarks	MiMo V2.6 Pro
Qwen3.8 Flash	90.5%	Alibaba	open	OpenRouter Benchmarks	Qwen3.8 Flash
Gemini 3 Flash Preview	90.4%	Google	undisclosed	OpenRouter Benchmarks	Gemini 3 Flash Preview
GPT-5.4	90.3%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.4
Claude Opus 5.5	90.2%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 5.5
Kimi K2.5	90.2%	Moonshot AI	open	OpenRouter Benchmarks	Kimi K2.5
Claude Opus 4.8	90.1%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 4.8
DeepSeek V4 Flash 0423	90.1%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V4 Flash 0423
Claude Fable 5.1	89.9%	Anthropic	closed	OpenRouter Benchmarks	Claude Fable 5.1
Claude Opus 5	89.9%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 5
Claude Opus 4.7	89.8%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 4.7
Kimi K2.6	89.7%	Moonshot AI	open	OpenRouter Benchmarks	Kimi K2.6
GLM-5.2	89.7%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM-5.2
MiniMax M3.1 Flash	89.3%	MiniMax	open	OpenRouter Benchmarks	MiniMax M3.1 Flash
GPT-5.6 Terra	89.3%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.6 Terra
GPT-6 Luna Pro	89%	OpenAI	undisclosed	OpenRouter Benchmarks	GPT-6 Luna Pro
Nova Micro 1.0	88.9%	Amazon	undisclosed	OpenRouter Benchmarks	Nova Micro 1.0
GPT-5.6 Luna	88.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.6 Luna
GPT-6 Luna	88.2%	OpenAI	closed	OpenRouter Benchmarks	GPT-6 Luna
Qwen3.5 397B A17B	88.1%	Alibaba	open	OpenRouter Benchmarks	Qwen3.5 397B A17B
GLM-5.3	87.9%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM-5.3
Qwen3.7 Plus	87.9%	Alibaba	closed	OpenRouter Benchmarks	Qwen3.7 Plus
Claude Fable 5	87.4%	Anthropic	closed	OpenRouter Benchmarks	Claude Fable 5
Qwen3.8 27B	87.2%	Alibaba	open	OpenRouter Benchmarks	Qwen3.8 27B
Claude Haiku 5.5	87.1%	Anthropic	closed	OpenRouter Benchmarks	Claude Haiku 5.5
GPT-5.1	87%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.1
Claude Sonnet 5	86.9%	Anthropic	closed	OpenRouter Benchmarks	Claude Sonnet 5
MiMo-V2.5-Pro	86.5%	Xiaomi	open	OpenRouter Benchmarks	MiMo-V2.5-Pro
Claude Opus 4.5	86.4%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 4.5
Claude Opus 4.6	86.4%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 4.6
MiniMax M2.7	86.1%	MiniMax	open	OpenRouter Benchmarks	MiniMax M2.7
Gemma 4 31B	85.9%	Google	open	OpenRouter Benchmarks	Gemma 4 31B
GPT-5	85.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-5
MiMo-V2.6-Flash	85.2%	Xiaomi	open	OpenRouter Benchmarks	MiMo-V2.6-Flash
Qwen3.6 27B	85.1%	Alibaba	open	OpenRouter Benchmarks	Qwen3.6 27B
Qwen3.6 35B A3B	85.1%	Alibaba	open	OpenRouter Benchmarks	Qwen3.6 35B A3B
MiniMax M2.5	85%	MiniMax	open	OpenRouter Benchmarks	MiniMax M2.5
MiMo-V2.5	84.9%	Xiaomi	open	OpenRouter Benchmarks	MiMo-V2.5
Ling 3.0 Flash VL	84.8%	Ant Group	open	OpenRouter Benchmarks	Ling 3.0 Flash VL
Qwen3.5-122B-A10B	84.8%	Alibaba	open	OpenRouter Benchmarks	Qwen3.5-122B-A10B
Gemini 3.5 Flash Lite	84.4%	Google	closed	OpenRouter Benchmarks	Gemini 3.5 Flash Lite
GLM 4.6	84.3%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 4.6
GPT-5.4 Mini	84.2%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.4 Mini
GLM 5.1	84.1%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 5.1
GLM 4.7	83.8%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 4.7
DeepSeek V3.2	83%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V3.2
Qwen3.5-35B-A3B	82.9%	Alibaba	open	OpenRouter Benchmarks	Qwen3.5-35B-A3B
Gemini 2.5 Pro	82.6%	Google	closed	OpenRouter Benchmarks	Gemini 2.5 Pro
DeepSeek V3.1 Terminus	82.4%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V3.1 Terminus
Gemini 3.1 Flash Lite	82.1%	Google	closed	OpenRouter Benchmarks	Gemini 3.1 Flash Lite
Muse Glimmer 30B	82.1%	Meta	open	OpenRouter Benchmarks	Muse Glimmer 30B
Claude Sonnet 4.5	81.7%	Anthropic	closed	OpenRouter Benchmarks	Claude Sonnet 4.5
DeepSeek V3.1	81.7%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V3.1
GPT-5 Mini	81%	OpenAI	closed	OpenRouter Benchmarks	GPT-5 Mini
Qwen3.5-9B	81%	Alibaba	open	OpenRouter Benchmarks	Qwen3.5-9B
GLM 5	81%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 5
DeepSeek R1	79.7%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek R1
Qwen3 235B A22B Thinking 2507	79.1%	Alibaba	open	OpenRouter Benchmarks	Qwen3 235B A22B Thinking 2507
Gemma 4 26B A4B	78.8%	Google	open	OpenRouter Benchmarks	Gemma 4 26B A4B
GPT-5.4 Nano	77.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.4 Nano
gpt-oss-120b	77.4%	OpenAI	open	OpenRouter Benchmarks	gpt-oss-120b
Qwen3-235B-A22B	77.4%	Alibaba	open	OpenRouter Benchmarks	Qwen3-235B-A22B
Gemini 2.5 Flash	76.7%	Google	closed	OpenRouter Benchmarks	Gemini 2.5 Flash
Ling 3.0 Flash	76.5%	Ant Group	open	OpenRouter Benchmarks	Ling 3.0 Flash
Qwen3 Coder Next	76.3%	Alibaba	open	OpenRouter Benchmarks	Qwen3 Coder Next
Mistral Small 4	75.8%	Mistral AI	open	OpenRouter Benchmarks	Mistral Small 4
Nemotron 3.5 Lightning	74.8%	NVIDIA	open	OpenRouter Benchmarks	Nemotron 3.5 Lightning
Claude Haiku 4.5	73%	Anthropic	closed	OpenRouter Benchmarks	Claude Haiku 4.5
Qwen3 Next 80B A3B Instruct	71.8%	Alibaba	open	OpenRouter Benchmarks	Qwen3 Next 80B A3B Instruct
Qwen3 VL 235B A22B Instruct	71.1%	Alibaba	open	OpenRouter Benchmarks	Qwen3 VL 235B A22B Instruct
Claude Sonnet 4.6	70.6%	Anthropic	closed	OpenRouter Benchmarks	Claude Sonnet 4.6
GPT-5 Nano	70.5%	OpenAI	closed	OpenRouter Benchmarks	GPT-5 Nano
Nemotron 3 Nano 30B A3B	68.8%	NVIDIA	open	OpenRouter Benchmarks	Nemotron 3 Nano 30B A3B
gpt-oss-20b	68.7%	OpenAI	open	OpenRouter Benchmarks	gpt-oss-20b
DeepSeek V3	66.9%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V3
Qwen3 VL 30B A3B Instruct	66.8%	Alibaba	open	OpenRouter Benchmarks	Qwen3 VL 30B A3B Instruct
GLM 4.6V	66.8%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 4.6V
DeepSeek V3	66.7%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V3
Llama 4 Maverick	66.7%	Meta	open	OpenRouter Benchmarks	Llama 4 Maverick
GPT-4.1	66%	OpenAI	closed	OpenRouter Benchmarks	GPT-4.1
Granite 4.2 8B	65.7%	IBM	open	OpenRouter Benchmarks	Granite 4.2 8B
GPT-4.1 Mini	65.2%	OpenAI	closed	OpenRouter Benchmarks	GPT-4.1 Mini
Qwen3 30B A3B Instruct 2507	64.8%	Alibaba	open	OpenRouter Benchmarks	Qwen3 30B A3B Instruct 2507
Qwen3 32B	63.3%	Alibaba	open	OpenRouter Benchmarks	Qwen3 32B
Qwen3-Coder	61.2%	Alibaba	open	OpenRouter Benchmarks	Qwen3-Coder
Qwen3 14B	61.2%	Alibaba	open	OpenRouter Benchmarks	Qwen3 14B
Qwen3 30B A3B	60.9%	Alibaba	open	OpenRouter Benchmarks	Qwen3 30B A3B
Gemini 2.5 Flash Lite	59.6%	Google	undisclosed	OpenRouter Benchmarks	Gemini 2.5 Flash Lite
GLM 4.7 Flash	58%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 4.7 Flash
Qwen3 VL 8B Instruct	53.9%	Alibaba	open	OpenRouter Benchmarks	Qwen3 VL 8B Instruct
Qwen3 Coder 30B A3B Instruct	53.4%	Alibaba	open	OpenRouter Benchmarks	Qwen3 Coder 30B A3B Instruct
GPT-4o	52.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-4o
GPT-4.1 Nano	50.5%	OpenAI	closed	OpenRouter Benchmarks	GPT-4.1 Nano
Llama 3.3 70B Instruct	49.8%	Meta	open	OpenRouter Benchmarks	Llama 3.3 70B Instruct
Qwen2.5 72B Instruct	47.7%	Alibaba	open	OpenRouter Benchmarks	Qwen2.5 72B Instruct
Qwen2.5 VL 72B Instruct	45.7%	Alibaba	open	OpenRouter Benchmarks	Qwen2.5 VL 72B Instruct
GPT-4o-mini	43.6%	OpenAI	closed	OpenRouter Benchmarks	GPT-4o-mini
Gemma 3 12B	40.6%	Google	open	OpenRouter Benchmarks	Gemma 3 12B
Hy-MT2-30B-A3B	38.9%	Tencent	open	OpenRouter Benchmarks	Hy-MT2-30B-A3B
Llama 3.1 8B Instruct	29%	Meta	open	OpenRouter Benchmarks	Llama 3.1 8B Instruct
Llama 3.2 3B Instruct	28.6%	Meta	open	OpenRouter Benchmarks	Llama 3.2 3B Instruct
Gemma 3 4B	27.1%	Google	open	OpenRouter Benchmarks	Gemma 3 4B
Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)	11.6%	Google	undisclosed	OpenRouter Benchmarks	Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)

τ-bench Verified (airline)

How to read each metric

Search modelLabWeights

τ-bench Verified (airline)
Model	%	Lab	Weights	Source	Compare models
Claude Fable 5	82%	Anthropic	closed	OpenRouter Benchmarks	Claude Fable 5
DeepSeek V4 Flash 0423	81.7%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V4 Flash 0423
Gemini 3.7 Flash	80.7%	Google	closed	OpenRouter Benchmarks	Gemini 3.7 Flash
Qwen3.5 397B A17B	80.2%	Alibaba	open	OpenRouter Benchmarks	Qwen3.5 397B A17B
GPT-5.1	80%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.1
Ling 3.1 Flash	79.3%	Ant Group	undisclosed	OpenRouter Benchmarks	Ling 3.1 Flash
Claude Fable 5.1	79.2%	Anthropic	closed	OpenRouter Benchmarks	Claude Fable 5.1
Qwen3.6 27B	79%	Alibaba	open	OpenRouter Benchmarks	Qwen3.6 27B
MiMo V2.6 Pro	78.9%	Xiaomi	open	OpenRouter Benchmarks	MiMo V2.6 Pro
Claude Sonnet 5.5	78.8%	Anthropic	closed	OpenRouter Benchmarks	Claude Sonnet 5.5
GLM-5.3	78.8%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM-5.3
Gemini 4 Argon	78.7%	Google	closed	OpenRouter Benchmarks	Gemini 4 Argon
Kimi K3.1 Pro	78.7%	Moonshot AI	open	OpenRouter Benchmarks	Kimi K3.1 Pro
Claude Opus 5	78.7%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 5
Qwen3.5-35B-A3B	78.7%	Alibaba	open	OpenRouter Benchmarks	Qwen3.5-35B-A3B
DeepSeek V4 Pro	78.6%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V4 Pro
Gemini 3.5 Flash Lite	78.3%	Google	closed	OpenRouter Benchmarks	Gemini 3.5 Flash Lite
MiMo-V2.6-Flash	78.2%	Xiaomi	open	OpenRouter Benchmarks	MiMo-V2.6-Flash
Qwen3.8 27B	78%	Alibaba	open	OpenRouter Benchmarks	Qwen3.8 27B
DeepSeek V4.1 Flash	77.8%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V4.1 Flash
Claude Opus 4.7	77.8%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 4.7
Qwen3.5-122B-A10B	77.7%	Alibaba	open	OpenRouter Benchmarks	Qwen3.5-122B-A10B
Hy4	77.6%	Tencent	open	OpenRouter Benchmarks	Hy4
Claude Opus 4.8	77.5%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 4.8
Qwen3.8 2.4T A95B	77.4%	Alibaba	open	OpenRouter Benchmarks	Qwen3.8 2.4T A95B
Kimi K2.7 Code	77.3%	Moonshot AI	open	OpenRouter Benchmarks	Kimi K2.7 Code
Muse Glimmer 30B	77.2%	Meta	open	OpenRouter Benchmarks	Muse Glimmer 30B
Gemini 3.1 Pro	77%	Google	closed	OpenRouter Benchmarks	Gemini 3.1 Pro
Gemma 4 31B	77%	Google	open	OpenRouter Benchmarks	Gemma 4 31B
GLM 5	76.9%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 5
Gemini 3.8 Flash	76.7%	Google	closed	OpenRouter Benchmarks	Gemini 3.8 Flash
GLM-5.2	76.7%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM-5.2
Nova Micro 1.0	76.7%	Amazon	undisclosed	OpenRouter Benchmarks	Nova Micro 1.0
DeepSeek V4 Flash Vision Exp	76.7%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V4 Flash Vision Exp
Claude Opus 4.5	76.6%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 4.5
Qwen3.8 Flash	76.4%	Alibaba	open	OpenRouter Benchmarks	Qwen3.8 Flash
GLM-5.3-Flash	76.3%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM-5.3-Flash
MiniMax M3	76.3%	MiniMax	open	OpenRouter Benchmarks	MiniMax M3
Gemini 3.6 Flash	76.3%	Google	closed	OpenRouter Benchmarks	Gemini 3.6 Flash
Kimi K3	76.2%	Moonshot AI	open	OpenRouter Benchmarks	Kimi K3
Qwen3.6 35B A3B	76.2%	Alibaba	open	OpenRouter Benchmarks	Qwen3.6 35B A3B
GLM 4.7	76%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 4.7
Claude Opus 4.6	75.8%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 4.6
GLM 5.1	75.8%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 5.1
Muse Spark Open	75.7%	Meta	open	OpenRouter Benchmarks	Muse Spark Open
Qwen3.5-9B	75.7%	Alibaba	open	OpenRouter Benchmarks	Qwen3.5-9B
Claude Sonnet 5	75.4%	Anthropic	closed	OpenRouter Benchmarks	Claude Sonnet 5
Claude Opus 5.5	75.3%	Anthropic	closed	OpenRouter Benchmarks	Claude Opus 5.5
Grok 4.7	75.3%	xAI	closed	OpenRouter Benchmarks	Grok 4.7
Grok 4.6	75.3%	xAI	closed	OpenRouter Benchmarks	Grok 4.6
GPT-5.3-Codex	75.3%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.3-Codex
GPT-5.4	75%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.4
Claude Sonnet 4.5	74.9%	Anthropic	closed	OpenRouter Benchmarks	Claude Sonnet 4.5
Kimi K2.6	74.9%	Moonshot AI	open	OpenRouter Benchmarks	Kimi K2.6
Nemotron 3.5 Super VL	74.9%	NVIDIA	open	OpenRouter Benchmarks	Nemotron 3.5 Super VL
Claude Sonnet 4.6	74.8%	Anthropic	closed	OpenRouter Benchmarks	Claude Sonnet 4.6
GPT-5.5	74.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.5
DeepSeek V3.2	74.7%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V3.2
MiniMax M3.1 Flash	74.7%	MiniMax	open	OpenRouter Benchmarks	MiniMax M3.1 Flash
MiMo-V2.5	74.7%	Xiaomi	open	OpenRouter Benchmarks	MiMo-V2.5
GPT-5.6 Sol	74.2%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.6 Sol
MiMo-V2.5-Pro	74.1%	Xiaomi	open	OpenRouter Benchmarks	MiMo-V2.5-Pro
Claude Haiku 5.5	74%	Anthropic	closed	OpenRouter Benchmarks	Claude Haiku 5.5
Ling 3.0 Flash	74%	Ant Group	open	OpenRouter Benchmarks	Ling 3.0 Flash
Kimi K2.5	73.8%	Moonshot AI	open	OpenRouter Benchmarks	Kimi K2.5
MiniMax M2.7	73.4%	MiniMax	open	OpenRouter Benchmarks	MiniMax M2.7
Gemini 3 Flash Preview	73.3%	Google	undisclosed	OpenRouter Benchmarks	Gemini 3 Flash Preview
GLM 4.6	73.3%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 4.6
DeepSeek V3.1	73.1%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V3.1
DeepSeek V3.1 Terminus	73%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V3.1 Terminus
Gemini 3.5 Flash	73%	Google	closed	OpenRouter Benchmarks	Gemini 3.5 Flash
Gemini 3.1 Flash Lite	72.7%	Google	closed	OpenRouter Benchmarks	Gemini 3.1 Flash Lite
GPT-5.6 Luna	72.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.6 Luna
GPT-6 Astra Pro	72.7%	OpenAI	undisclosed	OpenRouter Benchmarks	GPT-6 Astra Pro
Qwen3.7 Plus	72.7%	Alibaba	closed	OpenRouter Benchmarks	Qwen3.7 Plus
GPT-5 Mini	72.4%	OpenAI	closed	OpenRouter Benchmarks	GPT-5 Mini
GPT-5.6 Terra	72.3%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.6 Terra
Ling 3.0 Flash VL	72%	Ant Group	open	OpenRouter Benchmarks	Ling 3.0 Flash VL
Qwen3.7 Max	72%	Alibaba	closed	OpenRouter Benchmarks	Qwen3.7 Max
GPT-5.6 Sol Pro	71.7%	OpenAI	undisclosed	OpenRouter Benchmarks	GPT-5.6 Sol Pro
Nemotron 3.5 Lightning	71.5%	NVIDIA	open	OpenRouter Benchmarks	Nemotron 3.5 Lightning
GPT-6.1 Sol	70.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-6.1 Sol
GPT-5	70.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-5
GPT-5.6 Luna Pro	70.7%	OpenAI	undisclosed	OpenRouter Benchmarks	GPT-5.6 Luna Pro
Gemma 4 26B A4B	70.6%	Google	open	OpenRouter Benchmarks	Gemma 4 26B A4B
GPT-6 Sol	70.2%	OpenAI	closed	OpenRouter Benchmarks	GPT-6 Sol
GPT-6.1 Terra	70%	OpenAI	closed	OpenRouter Benchmarks	GPT-6.1 Terra
GPT-6 Astra	69.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-6 Astra
GLM 4.7 Flash	69.7%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 4.7 Flash
GPT-6 Luna Pro	69.3%	OpenAI	undisclosed	OpenRouter Benchmarks	GPT-6 Luna Pro
MiniMax M2.5	69.2%	MiniMax	open	OpenRouter Benchmarks	MiniMax M2.5
GPT-6 Luna	68.4%	OpenAI	closed	OpenRouter Benchmarks	GPT-6 Luna
gpt-oss-120b	67.3%	OpenAI	open	OpenRouter Benchmarks	gpt-oss-120b
GPT-5.4 Mini	66.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.4 Mini
Claude Haiku 4.5	66.4%	Anthropic	closed	OpenRouter Benchmarks	Claude Haiku 4.5
GPT-5.4 Nano	65.6%	OpenAI	closed	OpenRouter Benchmarks	GPT-5.4 Nano
Granite 4.2 8B	64%	IBM	open	OpenRouter Benchmarks	Granite 4.2 8B
Qwen3 235B A22B Thinking 2507	61.7%	Alibaba	open	OpenRouter Benchmarks	Qwen3 235B A22B Thinking 2507
Qwen3 Coder Next	60.7%	Alibaba	open	OpenRouter Benchmarks	Qwen3 Coder Next
GLM 4.6V	60.7%	Zhipu AI (Z.ai)	open	OpenRouter Benchmarks	GLM 4.6V
Gemini 2.5 Pro	60%	Google	closed	OpenRouter Benchmarks	Gemini 2.5 Pro
Gemini 2.5 Flash	59.1%	Google	closed	OpenRouter Benchmarks	Gemini 2.5 Flash
DeepSeek R1	58%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek R1
Qwen3 32B	54.3%	Alibaba	open	OpenRouter Benchmarks	Qwen3 32B
Gemini 2.5 Flash Lite	53.7%	Google	undisclosed	OpenRouter Benchmarks	Gemini 2.5 Flash Lite
gpt-oss-20b	53.5%	OpenAI	open	OpenRouter Benchmarks	gpt-oss-20b
GPT-4.1	52.3%	OpenAI	closed	OpenRouter Benchmarks	GPT-4.1
Nemotron 3 Nano 30B A3B	52%	NVIDIA	open	OpenRouter Benchmarks	Nemotron 3 Nano 30B A3B
GPT-5 Nano	48.3%	OpenAI	closed	OpenRouter Benchmarks	GPT-5 Nano
Qwen3-Coder	48%	Alibaba	open	OpenRouter Benchmarks	Qwen3-Coder
Gemma 3 12B	48%	Google	open	OpenRouter Benchmarks	Gemma 3 12B
Qwen3 VL 235B A22B Instruct	48%	Alibaba	open	OpenRouter Benchmarks	Qwen3 VL 235B A22B Instruct
Qwen3-235B-A22B	47.9%	Alibaba	open	OpenRouter Benchmarks	Qwen3-235B-A22B
Qwen3 Next 80B A3B Instruct	47.5%	Alibaba	open	OpenRouter Benchmarks	Qwen3 Next 80B A3B Instruct
Qwen3 30B A3B	47%	Alibaba	open	OpenRouter Benchmarks	Qwen3 30B A3B
Mistral Small 4	46%	Mistral AI	open	OpenRouter Benchmarks	Mistral Small 4
Qwen3 Coder 30B A3B Instruct	46%	Alibaba	open	OpenRouter Benchmarks	Qwen3 Coder 30B A3B Instruct
GPT-4.1 Mini	45.1%	OpenAI	closed	OpenRouter Benchmarks	GPT-4.1 Mini
Qwen3 14B	44%	Alibaba	open	OpenRouter Benchmarks	Qwen3 14B
DeepSeek V3	43.8%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V3
Llama 4 Maverick	43.4%	Meta	open	OpenRouter Benchmarks	Llama 4 Maverick
GPT-4o	43.2%	OpenAI	closed	OpenRouter Benchmarks	GPT-4o
DeepSeek V3	43%	DeepSeek	open	OpenRouter Benchmarks	DeepSeek V3
Qwen3 30B A3B Instruct 2507	41.5%	Alibaba	open	OpenRouter Benchmarks	Qwen3 30B A3B Instruct 2507
Qwen3 VL 30B A3B Instruct	41.3%	Alibaba	open	OpenRouter Benchmarks	Qwen3 VL 30B A3B Instruct
Llama 3.3 70B Instruct	39.9%	Meta	open	OpenRouter Benchmarks	Llama 3.3 70B Instruct
Qwen3 VL 8B Instruct	32.7%	Alibaba	open	OpenRouter Benchmarks	Qwen3 VL 8B Instruct
GPT-4o-mini	30.7%	OpenAI	closed	OpenRouter Benchmarks	GPT-4o-mini
Qwen2.5 72B Instruct	27.3%	Alibaba	open	OpenRouter Benchmarks	Qwen2.5 72B Instruct
Llama 3.1 8B Instruct	23.5%	Meta	open	OpenRouter Benchmarks	Llama 3.1 8B Instruct
GPT-4.1 Nano	14%	OpenAI	closed	OpenRouter Benchmarks	GPT-4.1 Nano

Model against model

Sources

Epoch AI, CC BY 4.0, latest version we checked.
OpenRouter Benchmarks, CC BY 4.0, latest version we checked.
OpenRouter, dados públicos da API; exports de benchmarks e rankings em CC BY 4.0, latest version we checked.

Dates come from the sources: OpenRouter listing or Hugging Face repo creation, which may differ from the official launch. Family and version were grouped by rule from the name.