Benchmark Explorer

Compare 470 AI models across four benchmark categories: reasoning, code generation, language understanding, and tool use. Scores are derived from model capabilities and composite rankings.

Total Models

470

With Reasoning

345

With Vision

307

Avg Score

Top Models by Average Benchmark Score

LMMarketCap.com

Benchmark Categories

Top models in each benchmark category, scored on capability support and composite ranking data.

Reasoning

Chain-of-thought reasoning, logic, and mathematical problem solving.

#	Model	Provider	Score
1	Claude Fable 5.1Anthropic	Anthropic	100
2	Claude Fable 5.1 (batch)Anthropic	Anthropic	100
3	Claude Fable 5Anthropic	Anthropic	100
4	Claude Fable 5 (batch)Anthropic	Anthropic	100
5	Claude Opus 5.5Anthropic	Anthropic	100
6	Claude Opus 5Anthropic	Anthropic	100
7	Claude Opus 4.8Anthropic	Anthropic	100
8	Claude Opus 4.7Anthropic	Anthropic	100
9	GPT-5.5 ProOpenAI	OpenAI	100
10	GPT-5.5 Pro (batch)OpenAI	OpenAI	100
11	GPT-5.5OpenAI	OpenAI	100
12	Gemini 3.1 Pro Preview Custom ToolsGoogle	Google	100
13	Gemini 3.1 Pro PreviewGoogle	Google	100
14	Gemini 3.1 Pro Preview (batch)Google	Google	100
15	GPT-5.4 ProOpenAI	OpenAI	100

Code Generation

Code writing, debugging, and software engineering tasks.

#	Model	Provider	Score
1	Claude Fable 5.1Anthropic	Anthropic	92
2	Claude Fable 5.1 (batch)Anthropic	Anthropic	92
3	Claude Fable 5Anthropic	Anthropic	92
4	Claude Fable 5 (batch)Anthropic	Anthropic	92
5	Claude Opus 5.5Anthropic	Anthropic	91
6	Claude Opus 5Anthropic	Anthropic	91
7	Claude Opus 4.8Anthropic	Anthropic	91
8	Claude Opus 4.7Anthropic	Anthropic	91
9	GPT-5.5 ProOpenAI	OpenAI	89
10	GPT-5.5 Pro (batch)OpenAI	OpenAI	89
11	GPT-5.5OpenAI	OpenAI	89
12	Gemini 3.1 Pro Preview Custom ToolsGoogle	Google	89
13	Gemini 3.1 Pro PreviewGoogle	Google	89
14	Gemini 3.1 Pro Preview (batch)Google	Google	89
15	GPT-5.4 ProOpenAI	OpenAI	89

Language Understanding

Text comprehension, multi-modal input, and information retrieval.

#	Model	Provider	Score
1	Claude Fable 5.1Anthropic	Anthropic	85
2	Claude Fable 5.1 (batch)Anthropic	Anthropic	85
3	Claude Fable 5Anthropic	Anthropic	85
4	Claude Fable 5 (batch)Anthropic	Anthropic	85
5	Claude Opus 5.5Anthropic	Anthropic	85
6	Claude Opus 5Anthropic	Anthropic	85
7	Claude Opus 4.8Anthropic	Anthropic	85
8	Claude Opus 4.7Anthropic	Anthropic	85
9	GPT-5.5 ProOpenAI	OpenAI	83
10	GPT-5.5 Pro (batch)OpenAI	OpenAI	83
11	GPT-5.5OpenAI	OpenAI	83
12	Gemini 3.1 Pro Preview Custom ToolsGoogle	Google	83
13	Gemini 3.1 Pro PreviewGoogle	Google	83
14	Gemini 3.1 Pro Preview (batch)Google	Google	83
15	GPT-5.4 ProOpenAI	OpenAI	82

Tool Use

Function calling, structured output, and agentic workflows.

#	Model	Provider	Score
1	Claude Fable 5.1Anthropic	Anthropic	98
2	Claude Fable 5.1 (batch)Anthropic	Anthropic	98
3	Claude Fable 5Anthropic	Anthropic	98
4	Claude Fable 5 (batch)Anthropic	Anthropic	98
5	Claude Opus 5.5Anthropic	Anthropic	98
6	Claude Opus 5Anthropic	Anthropic	98
7	Claude Opus 4.8Anthropic	Anthropic	98
8	Claude Opus 4.7Anthropic	Anthropic	98
9	GPT-5.5 ProOpenAI	OpenAI	98
10	GPT-5.5 Pro (batch)OpenAI	OpenAI	98
11	GPT-5.5OpenAI	OpenAI	98
12	Gemini 3.1 Pro Preview Custom ToolsGoogle	Google	98
13	Gemini 3.1 Pro PreviewGoogle	Google	98
14	Gemini 3.1 Pro Preview (batch)Google	Google	98
15	GPT-5.4 ProOpenAI	OpenAI	98

Cross-Benchmark Comparison

All 470 models ranked by average score across all four benchmark categories.

Model	Provider	Code	Reasoning	Language	Tool Use	Average
Claude Fable 5.1Anthropic	Anthropic	92	100	85	98	94
Claude Fable 5.1 (batch)Anthropic	Anthropic	92	100	85	98	94
Claude Fable 5Anthropic	Anthropic	92	100	85	98	94
Claude Fable 5 (batch)Anthropic	Anthropic	92	100	85	98	94
Claude Opus 5.5Anthropic	Anthropic	91	100	85	98	94
Claude Opus 5Anthropic	Anthropic	91	100	85	98	94
Claude Opus 4.8Anthropic	Anthropic	91	100	85	98	94
Claude Opus 4.7Anthropic	Anthropic	91	100	85	98	94
GPT-5.5 ProOpenAI	OpenAI	89	100	83	98	93
GPT-5.5 Pro (batch)OpenAI	OpenAI	89	100	83	98	93
GPT-5.5OpenAI	OpenAI	89	100	83	98	93
Gemini 3.1 Pro Preview Custom ToolsGoogle	Google	89	100	83	98	93
Gemini 3.1 Pro PreviewGoogle	Google	89	100	83	98	93
Gemini 3.1 Pro Preview (batch)Google	Google	89	100	83	98	93
GPT-5.4 ProOpenAI	OpenAI	89	100	82	98	92
GPT-5.4 Pro (batch)OpenAI	OpenAI	89	100	82	98	92
GPT-5.4OpenAI	OpenAI	89	100	82	98	92
Claude Opus 4.6Anthropic	Anthropic	87	100	81	98	92
GPT-5.6 Luna ProOpenAI	OpenAI	86	100	80	98	91
GPT-5.6 Luna Pro (batch)OpenAI	OpenAI	86	100	80	98	91
GPT-5.6 LunaOpenAI	OpenAI	86	100	80	98	91
GPT-5.6 Luna (batch)OpenAI	OpenAI	86	100	80	98	91
GPT-5.6 Terra ProOpenAI	OpenAI	86	100	80	98	91
GPT-5.6 Terra Pro (batch)OpenAI	OpenAI	86	100	80	98	91
GPT-5.6 TerraOpenAI	OpenAI	86	100	80	98	91
GPT-5.6 Terra (batch)OpenAI	OpenAI	86	100	80	98	91
GPT-5.6 Sol ProOpenAI	OpenAI	86	100	80	98	91
GPT-5.6 Sol Pro (batch)OpenAI	OpenAI	86	100	80	98	91
GPT-5.6 SolOpenAI	OpenAI	86	100	80	98	91
GPT-5.6 Sol (batch)OpenAI	OpenAI	86	100	80	98	91
Grok 4.3xAI	xAI	86	100	80	98	91
Grok 4.3 (batch)xAI	xAI	86	100	80	98	91
GPT-5.5 (batch)OpenAI	OpenAI	86	100	80	98	91
Grok 4.20xAI	xAI	86	100	80	98	91
GPT-5.4 (batch)OpenAI	OpenAI	86	100	80	98	91
GPT-5.3-CodexOpenAI	OpenAI	87	98	81	98	91
GPT-5.2-CodexOpenAI	OpenAI	87	98	81	98	91
Gemini 3 Flash PreviewGoogle	Google	86	100	80	98	91
Gemini 3 Flash Preview (batch)Google	Google	86	100	80	98	91
GPT-5.2 ProOpenAI	OpenAI	87	98	81	98	91
GPT-5.2 Pro (batch)OpenAI	OpenAI	87	98	81	98	91
GPT-5.2OpenAI	OpenAI	87	98	81	98	91
Grok 4.7xAI	xAI	86	98	80	98	91
Grok 4.7 (batch)xAI	xAI	86	98	80	98	91
Grok 4.6xAI	xAI	86	98	80	98	91
Grok 4.6 (batch)xAI	xAI	86	98	80	98	91
Grok 4.5xAI	xAI	86	98	80	98	91
GPT-5.2 (batch)OpenAI	OpenAI	86	98	80	98	91
GPT-5.1-Codex-MaxOpenAI	OpenAI	86	98	80	98	91
GPT-5.1OpenAI	OpenAI	86	98	80	98	91
GPT-5.1 (batch)OpenAI	OpenAI	86	98	80	98	91
GPT-5.1-CodexOpenAI	OpenAI	86	98	80	98	91
GPT-5.1-Codex-MiniOpenAI	OpenAI	86	98	80	98	91
GPT-5 ProOpenAI	OpenAI	86	98	80	98	91
GPT-5 Pro (batch)OpenAI	OpenAI	86	98	80	98	91
GPT-5OpenAI	OpenAI	86	98	80	98	91
GPT-5 (batch)OpenAI	OpenAI	86	98	80	98	91
Claude Sonnet 5.5Anthropic	Anthropic	83	100	78	98	90
Claude Sonnet 5Anthropic	Anthropic	83	100	78	98	90
Claude Sonnet 4.6Anthropic	Anthropic	83	100	78	98	90
o3 ProOpenAI	OpenAI	84	98	79	98	90
o3OpenAI	OpenAI	84	98	79	98	90
o3 (batch)OpenAI	OpenAI	84	98	79	98	90
Claude Opus 4.5Anthropic	Anthropic	83	98	78	98	89
Claude Opus 4.5 (batch)Anthropic	Anthropic	83	98	78	98	89
Claude Opus 4.8 (batch)Anthropic	Anthropic	82	100	76	98	89
Claude Opus 4.7 (batch)Anthropic	Anthropic	82	100	76	98	89
Claude Opus 4.6 (batch)Anthropic	Anthropic	82	100	76	98	89
Gemini 2.5 ProGoogle	Google	82	100	76	98	89
Gemini 2.5 Pro Preview 06-05Google	Google	82	100	76	98	89
Claude Sonnet 4.5Anthropic	Anthropic	81	100	76	98	89
GPT-6 AstraOpenAI	OpenAI	80	100	75	98	88
GPT-6 Astra (batch)OpenAI	OpenAI	80	100	75	98	88
GPT-6 Astra ProOpenAI	OpenAI	80	100	75	98	88
GPT-6 Astra Pro (batch)OpenAI	OpenAI	80	100	75	98	88
Gemini 3.8 FlashGoogle	Google	80	100	75	98	88
Gemini 3.8 Flash (batch)Google	Google	80	100	75	98	88
Claude Opus 4.1 (batch)Anthropic	Anthropic	81	98	76	98	88
o4 Mini (batch)OpenAI	OpenAI	80	98	75	98	88
Gemini 3.5 Flash LiteGoogle	Google	78	100	73	98	87
Gemini 3.5 Flash Lite (batch)Google	Google	78	100	73	98	87
Gemini 3.5 FlashGoogle	Google	78	100	73	98	87
Gemini 3.5 Flash (batch)Google	Google	78	100	73	98	87
Gemini 2.5 Flash LiteGoogle	Google	78	100	73	98	87
Gemini 2.5 FlashGoogle	Google	78	100	73	98	87
GPT-5.4 NanoOpenAI	OpenAI	78	98	74	98	87
GPT-5.4 Nano (batch)OpenAI	OpenAI	78	98	74	98	87
GPT-5.4 MiniOpenAI	OpenAI	78	98	74	98	87
GPT-5.4 Mini (batch)OpenAI	OpenAI	78	98	74	98	87
Claude Sonnet 4.6 (batch)Anthropic	Anthropic	75	100	71	98	86
Claude Sonnet 4.5 (batch)Anthropic	Anthropic	75	100	71	98	86
GPT-5 Mini (batch)OpenAI	OpenAI	76	98	72	98	86
GPT-5 NanoOpenAI	OpenAI	76	98	72	98	86
GPT-5 Nano (batch)OpenAI	OpenAI	76	98	72	98	86
Claude Opus 4.1Anthropic	Anthropic	76	98	75	90	85
Gemini 3.1 Flash Lite PreviewGoogle	Google	72	100	67	98	84
Claude Haiku 5.5Anthropic	Anthropic	71	100	67	98	84
DeepSeek V3.2DeepSeek	DeepSeek	82	98	58	98	84
Gemma 4 31BGoogle	Google	76	98	63	98	84
Gemma 4 31B (free)Google	Google	76	98	63	98	84
Claude Haiku 4.5Anthropic	Anthropic	71	98	67	98	84
Claude Haiku 4.5 (batch)Anthropic	Anthropic	71	98	67	98	84
R1 0528DeepSeek	DeepSeek	79	98	56	98	83
o3 Mini (batch)OpenAI	OpenAI	74	98	60	98	83
GLM 5.3 FlashX LegalZhipu AI	Zhipu AI	71	100	59	98	82
GLM 5.3 FlashXZhipu AI	Zhipu AI	71	100	59	98	82
GLM 5.3 FlashZhipu AI	Zhipu AI	71	100	59	98	82
GLM 5.3 Flash (batch)Zhipu AI	Zhipu AI	71	100	59	98	82
Qwen3.5 397B A17BAlibaba	Alibaba	72	98	59	98	82
Mistral Large 4Mistral AI	Mistral AI	70	100	58	98	82
Qwen3.5-122B-A10BAlibaba	Alibaba	71	98	59	98	82
Qwen3.5-27BAlibaba	Alibaba	70	98	59	98	81
GLM 5V TurboZhipu AI	Zhipu AI	70	98	58	98	81
Qwen3.5-35B-A3BAlibaba	Alibaba	70	98	58	98	81
Qwen3.5-9BAlibaba	Alibaba	68	98	57	98	80
Claude Haiku 5.5 (batch)Anthropic	Anthropic	61	100	59	98	80
DeepSeek V3.2 ExpDeepSeek	DeepSeek	72	98	50	98	80
DeepSeek V3.1DeepSeek	DeepSeek	72	98	50	98	80
GLM 4.5VZhipu AI	Zhipu AI	69	93	57	98	79
GLM 5.3 PrimeZhipu AI	Zhipu AI	70	100	48	98	79
GLM 5.3Zhipu AI	Zhipu AI	70	100	48	98	79
GLM 5.3 (batch)Zhipu AI	Zhipu AI	70	100	48	98	79
GLM 5.2Zhipu AI	Zhipu AI	70	100	48	98	79
GLM 5.1Zhipu AI	Zhipu AI	70	98	48	98	79
GLM 5 TurboZhipu AI	Zhipu AI	70	98	48	98	79
GLM 5Zhipu AI	Zhipu AI	70	98	48	98	79
GLM 4.7Zhipu AI	Zhipu AI	69	98	47	98	78
GLM 4.6Zhipu AI	Zhipu AI	69	98	47	98	78
GLM 4.5Zhipu AI	Zhipu AI	69	98	47	98	78
Grok 4.20 Multi-AgentxAI	xAI	85	100	80	43	77
MiniMax M2.7MiniMax	MiniMax	66	98	45	98	77
MiniMax M2.5MiniMax	MiniMax	66	98	45	98	77
GPT-5.2 ChatOpenAI	OpenAI	77	48	81	98	76
Command A+Cohere	Cohere	56	98	46	98	75
Composer 2Cursor	Cursor	63	98	46	90	74
Composer 2 FastCursor	Cursor	63	98	46	90	74
Claude Sonnet 5 (batch)Anthropic	Anthropic	50	100	48	98	74
Fugu Ultrasakana	sakana	50	100	48	98	74
Claude Fable Latest~anthropic	~anthropic	50	100	48	98	74
Gemini 3.1 Flash LiteGoogle	Google	50	100	48	98	74
Gemini 3.1 Flash Lite (batch)Google	Google	50	100	48	98	74
Claude Haiku Latest~anthropic	~anthropic	50	100	48	98	74
Gemini Pro Latest~google	~google	50	100	48	98	74
Gemini Flash Latest~google	~google	50	100	48	98	74
Claude Sonnet Latest~anthropic	~anthropic	50	100	48	98	74
Claude Opus Latest~anthropic	~anthropic	50	100	48	98	74
MiniMax M1MiniMax	MiniMax	61	100	45	90	74
Qwen3 235B A22B Thinking 2507Alibaba	Alibaba	62	98	45	90	74
Nano Banana Pro (Gemini 3 Pro Image)Google	Google	50	98	48	98	74
Grok Build 0.1xAI	xAI	50	98	48	98	74
GPT Mini Latest~openai	~openai	50	98	48	98	74
GLM 4.5 AirZhipu AI	Zhipu AI	61	98	45	90	74
Kimi K2.6Moonshot AI	Moonshot AI	53	98	43	98	73
Kimi K2.5Moonshot AI	Moonshot AI	53	98	43	98	73
GPT-6.1 Terra ProOpenAI	OpenAI	47	100	46	98	73
GPT-6.1 Terra Pro (batch)OpenAI	OpenAI	47	100	46	98	73
GPT-6.1 TerraOpenAI	OpenAI	47	100	46	98	73
GPT-6.1 Terra (batch)OpenAI	OpenAI	47	100	46	98	73
Gemini 4 ArgonGoogle	Google	47	100	46	98	73
Gemini 4 Argon (batch)Google	Google	47	100	46	98	73
Muse Wondermeta	meta	47	100	46	98	73
GPT-6.1 Sol ProOpenAI	OpenAI	47	100	46	98	73
GPT-6.1 Sol Pro (batch)OpenAI	OpenAI	47	100	46	98	73
GPT-6.1 SolOpenAI	OpenAI	47	100	46	98	73
GPT-6.1 Sol (batch)OpenAI	OpenAI	47	100	46	98	73
Claude Sonnet 5.5 (batch)Anthropic	Anthropic	47	100	46	98	73
GPT-6 Luna ProOpenAI	OpenAI	47	100	46	98	73
GPT-6 Luna Pro (batch)OpenAI	OpenAI	47	100	46	98	73
GPT-6 LunaOpenAI	OpenAI	47	100	46	98	73
GPT-6 Luna (batch)OpenAI	OpenAI	47	100	46	98	73
GPT-6 Sol ProOpenAI	OpenAI	47	100	46	98	73
GPT-6 Sol Pro (batch)OpenAI	OpenAI	47	100	46	98	73
GPT-6 SolOpenAI	OpenAI	47	100	46	98	73
GPT-6 Sol (batch)OpenAI	OpenAI	47	100	46	98	73
Claude Opus 5.5 (batch)Anthropic	Anthropic	47	100	46	98	73
GPT Astra Latest~openai	~openai	47	100	46	98	73
GPT Sol Latest~openai	~openai	47	100	46	98	73
GPT Terra Latest~openai	~openai	47	100	46	98	73
GPT Luna Latest~openai	~openai	47	100	46	98	73
Fugu Ultra v2sakana	sakana	47	100	46	98	73
Fugu Maxsakana	sakana	47	100	46	98	73
Muse Spark 1.3 Contributormeta	meta	47	100	46	98	73
Muse Spark 1.3meta	meta	47	100	46	98	73
Muse Spark 1.2 Contributormeta	meta	47	100	46	98	73
Gemini 3.7 FlashGoogle	Google	47	100	46	98	73
Gemini 3.7 Flash (batch)Google	Google	47	100	46	98	73
Muse Spark 1.2meta	meta	47	100	46	98	73
Claude Opus 5 (batch)Anthropic	Anthropic	47	100	46	98	73
Gemini 3.6 FlashGoogle	Google	47	100	46	98	73
Gemini 3.6 Flash (batch)Google	Google	47	100	46	98	73
Muse Spark 1.1meta	meta	47	100	46	98	73
Nano Banana 2.1Google	Google	50	93	48	98	72
Sakana Namazusakana	sakana	47	98	46	98	72
Grok Latest~x-ai	~x-ai	47	98	46	98	72
GPT-5 MiniOpenAI	OpenAI	47	98	46	98	72
Crouton Alphastealth	stealth	50	100	40	98	72
Qwen3.7 PlusAlibaba	Alibaba	50	100	40	98	72
MiniMax M3MiniMax	MiniMax	50	100	40	98	72
Kimi Latest~moonshotai	~moonshotai	50	100	40	98	72
Qwen3.5 Plus 2026-04-20Alibaba	Alibaba	50	100	40	98	72
Qwen3.6 FlashAlibaba	Alibaba	50	100	40	98	72
MiMo-V2.5Xiaomi	Xiaomi	50	100	40	98	72
Qwen3.6 PlusAlibaba	Alibaba	50	100	40	98	72
Qwen3.5-FlashAlibaba	Alibaba	50	100	40	98	72
Qwen3.5 Plus 2026-02-15Alibaba	Alibaba	50	100	40	98	72
Kimi K2.7 CodeMoonshot AI	Moonshot AI	50	98	40	98	72
Qwen3.7 Plus (batch)Alibaba	Alibaba	50	98	40	98	72
MiniMax M3 (batch)MiniMax	MiniMax	50	98	40	98	72
Step 3.7 FlashStepFun	StepFun	50	98	40	98	72
Mistral Medium 3.5Mistral AI	Mistral AI	50	98	40	98	72
Mistral Medium 3.5 (batch)Mistral AI	Mistral AI	50	98	40	98	72
Qwen3.6 35B A3BAlibaba	Alibaba	50	98	40	98	72
Qwen3.6 27BAlibaba	Alibaba	50	98	40	98	72
Gemma 4 26B A4B Google	Google	50	98	40	98	72
Gemma 4 26B A4B (free)Google	Google	50	98	40	98	72
Mistral Small 4Mistral AI	Mistral AI	50	98	40	98	72
Seed-2.0-LiteByteDance	ByteDance	50	98	40	98	72
Seed-2.0-MiniByteDance	ByteDance	50	98	40	98	72
Seed 1.6 FlashByteDance	ByteDance	50	98	40	98	72
Seed 1.6ByteDance	ByteDance	50	98	40	98	72
GLM 4.6VZhipu AI	Zhipu AI	50	98	40	98	72
Qwen3 VL 30B A3B ThinkingAlibaba	Alibaba	50	98	40	98	72
Hy4Tencent	Tencent	47	100	38	98	71
Particle-1particle-ai	particle-ai	47	100	38	98	71
Kimi K3.1 ProMoonshot AI	Moonshot AI	47	100	38	98	71
MiniMax M3.1 FlashMiniMax	MiniMax	47	100	38	98	71
DeepSeek V4.1 Flash 1017DeepSeek	DeepSeek	47	100	38	98	71
Muse Spark Openmeta	meta	47	100	38	98	71
Nemotron 3.5 Super VLNVIDIA	NVIDIA	47	100	38	98	71
Nemotron 3.5 Super VL (free)NVIDIA	NVIDIA	47	100	38	98	71
Step 5 PreviewStepFun	StepFun	47	100	38	98	71
Ember-1fireworks	fireworks	47	100	38	98	71
Qwen3.8 Max PrimeAlibaba	Alibaba	47	100	38	98	71
MiMo-V2.6-Pro-UltraSpeedXiaomi	Xiaomi	47	100	38	98	71
MiMo-V2.6-FlashXiaomi	Xiaomi	47	100	38	98	71
MiMo-V2.6-ProXiaomi	Xiaomi	47	100	38	98	71
Qwen3.8 Omni FlashAlibaba	Alibaba	47	100	38	98	71
DeepSeek Flash Latest~deepseek	~deepseek	47	100	38	98	71
DeepSeek V4.1 FlashDeepSeek	DeepSeek	47	100	38	98	71
DeepSeek V4.1 Flash (batch)DeepSeek	DeepSeek	47	100	38	98	71
Qwen3.8 Max (0902)Alibaba	Alibaba	47	100	38	98	71
GLM Flash Latest~z-ai	~z-ai	47	100	38	98	71
Qwen3.8 FlashAlibaba	Alibaba	47	100	38	98	71
DeepSeek V4 Flash Vision ExpDeepSeek	DeepSeek	47	100	38	98	71
Qwen3.8 27BAlibaba	Alibaba	47	100	38	98	71
Qwen3.7 FlashAlibaba	Alibaba	47	100	38	98	71
Kimi K3Moonshot AI	Moonshot AI	47	100	38	98	71
Kimi K3 (batch)Moonshot AI	Moonshot AI	47	100	38	98	71
Jev Routertypesafe	typesafe	47	100	38	98	71
Glimpseelorian	elorian	47	98	38	98	70
Ternary Bonsai 2 9Bprism-ml	prism-ml	47	98	38	98	70
Ternary Bonsai 2 27Bprism-ml	prism-ml	47	98	38	98	70
Ling 3.0 Flash VLinclusionai	inclusionai	47	98	38	98	70
Qwen3.8 27B (batch)Alibaba	Alibaba	47	98	38	98	70
Dots3-Note Preview (free)dots-studio	dots-studio	47	98	38	98	70
Seed 2.1 TurboByteDance	ByteDance	47	98	38	98	70
Seed-2.0-CodeByteDance	ByteDance	47	98	38	98	70
Muse Glimmer 30Bmeta	meta	47	98	38	98	70
Muse Glimmer 30B (batch)meta	meta	47	98	38	98	70
Qwen3.7 MaxAlibaba	Alibaba	50	100	30	98	70
DeepSeek V4 Pro 0423DeepSeek	DeepSeek	50	100	30	98	70
DeepSeek V4 Flash 0423DeepSeek	DeepSeek	50	100	30	98	70
MiMo-V2.5-ProXiaomi	Xiaomi	50	100	30	98	70
Nemotron 3 SuperNVIDIA	NVIDIA	50	100	30	98	70
GPT-4o (batch)OpenAI	OpenAI	63	48	68	98	69
Nemotron 3 UltraNVIDIA	NVIDIA	50	98	30	98	69
Mercury 2Inception	Inception	50	98	30	98	69
Aion-2.0aion-labs	aion-labs	50	98	30	98	69
GLM 4.7 FlashZhipu AI	Zhipu AI	50	98	30	98	69
Nemotron 3 Nano 30B A3BNVIDIA	NVIDIA	50	98	30	98	69
Kimi K2 ThinkingMoonshot AI	Moonshot AI	50	98	30	98	69
gpt-oss-safeguard-20bOpenAI	OpenAI	50	98	30	98	69
DeepSeek V3.1 TerminusDeepSeek	DeepSeek	50	98	30	98	69
gpt-oss-120bOpenAI	OpenAI	50	98	30	98	69
gpt-oss-120b (batch)OpenAI	OpenAI	50	98	30	98	69
gpt-oss-20bOpenAI	OpenAI	50	98	30	98	69
gpt-oss-20b (batch)OpenAI	OpenAI	50	98	30	98	69
Qwen3 32BAlibaba	Alibaba	50	98	30	98	69
Nova 2 LiteAmazon	Amazon	45	100	40	90	69
GPT-4.1OpenAI	OpenAI	59	53	65	98	69
GPT-4.1 (batch)OpenAI	OpenAI	59	53	65	98	69
Beam 1.0 501B A23Breflection	reflection	47	100	28	98	68
Mercury 3Inception	Inception	47	100	28	98	68
Ling 3.1 Flashinclusionai	inclusionai	47	100	28	98	68
Perceptron Mk1.5perceptron	perceptron	47	90	38	98	68
DeepSeek Pro Latest~deepseek	~deepseek	47	100	28	98	68
Hy4 previewTencent	Tencent	47	100	28	98	68
GLM Latest~z-ai	~z-ai	47	100	28	98	68
Qwen3.8 2.4T A95BAlibaba	Alibaba	47	100	28	98	68
DeepSeek V4 Pro 0813DeepSeek	DeepSeek	47	100	28	98	68
DeepSeek V4 Flash Latest~deepseek	~deepseek	47	100	28	98	68
DeepSeek V4 Flash 0731DeepSeek	DeepSeek	47	100	28	98	68
DeepSeek V4 Flash 0731 (batch)DeepSeek	DeepSeek	47	100	28	98	68
Nemotron 3 Nano Omni (free)NVIDIA	NVIDIA	45	98	40	90	68
Flint (free)recursive	recursive	47	98	28	98	68
Xing4.0 29B A4B (free)xingchen-agi	xingchen-agi	47	98	28	98	68
Aion 3.5 Miniaion-labs	aion-labs	47	98	28	98	68
Aion 3.5aion-labs	aion-labs	47	98	28	98	68
Solar Mini 4Upstage	Upstage	47	98	28	98	68
Mercury 2.5Inception	Inception	47	98	28	98	68
Granite 4.2 8BIBM	IBM	47	98	28	98	68
Nemotron 3.5 LightningNVIDIA	NVIDIA	47	98	28	98	68
Solar Pro 4Upstage	Upstage	47	98	28	98	68
Ling 3.0 Tinyinclusionai	inclusionai	47	98	28	98	68
Ling 3.0 Flashinclusionai	inclusionai	47	98	28	98	68
Aion-3.0-Miniaion-labs	aion-labs	47	98	28	98	68
Aion-3.0aion-labs	aion-labs	47	98	28	98	68
Hy3Tencent	Tencent	47	98	28	98	68
Qwen3 30B A3BAlibaba	Alibaba	50	93	30	98	68
Qwen3 14BAlibaba	Alibaba	50	93	30	98	68
LFM3-1.7BLiquid AI	Liquid AI	47	95	28	98	67
LFM3-1.7B (free)Liquid AI	Liquid AI	47	95	28	98	67
LFM2.5-24B-A2BLiquid AI	Liquid AI	47	95	28	98	67
LFM2.5-2.6BLiquid AI	Liquid AI	47	95	28	98	67
Inkling Smallthinkingmachines	thinkingmachines	42	98	38	90	67
Inkling Small (batch)thinkingmachines	thinkingmachines	42	98	38	90	67
Inklingthinkingmachines	thinkingmachines	42	98	38	90	67
GPT-4o (2024-11-20)OpenAI	OpenAI	62	48	60	98	67
GPT-4o (2024-08-06)OpenAI	OpenAI	62	48	60	98	67
GPT-4oOpenAI	OpenAI	62	48	60	98	67
Llama 4 MaverickMeta	Meta	61	48	59	98	67
Nemotron 3 Ultra (free)NVIDIA	NVIDIA	45	100	30	90	66
GPT-4o-miniOpenAI	OpenAI	60	48	59	98	66
GPT-4o (2024-05-13)OpenAI	OpenAI	62	45	60	98	66
LFM2.5-2.6B (free)Liquid AI	Liquid AI	47	90	28	98	66
North Mini Code (free)Cohere	Cohere	45	98	30	90	66
Hy3 previewTencent	Tencent	45	98	30	90	66
Step 3.5 FlashStepFun	StepFun	45	98	30	90	66
GPT-4o-mini (batch)OpenAI	OpenAI	55	48	62	98	66
Mistral Large 3 2512Mistral AI	Mistral AI	59	48	57	98	66
GPT-4.1 Mini (batch)OpenAI	OpenAI	52	53	59	98	66
GPT-4.1 Nano (batch)OpenAI	OpenAI	52	53	59	98	66
Nemotron 3.5 Lightning (free)NVIDIA	NVIDIA	42	100	28	90	65
LongCat 2.0Meituan	Meituan	42	100	28	90	65
Falcon-H1-Arabic 34B InstructTII	TII	45	95	30	90	65
Falcon-H1-Arabic 7B InstructTII	TII	45	95	30	90	65
Falcon-H1-Arabic 3B InstructTII	TII	45	95	30	90	65
DeepSeek V3 0324DeepSeek	DeepSeek	62	48	50	98	65
Llama 4 ScoutMeta	Meta	53	53	52	98	64
DeepSeek V3DeepSeek	DeepSeek	61	48	49	98	64
SWE-1.5Windsurf	Windsurf	45	90	30	90	64
Falcon Arabic 7B InstructTII	TII	45	90	30	90	64
Falcon3 10B InstructTII	TII	45	90	30	90	64
Falcon3 7B InstructTII	TII	45	90	30	90	64
GPT-4 Turbo (batch)OpenAI	OpenAI	51	45	58	98	63
Llama 3.3 70B InstructMeta	Meta	58	48	47	98	63
Mistral Large 2407Mistral AI	Mistral AI	58	48	46	98	63
Mistral LargeMistral AI	Mistral AI	58	48	46	98	63
Qwen3 235B A22B Instruct 2507Alibaba	Alibaba	57	48	45	98	62
Mixtral 8x22B InstructMistral AI	Mistral AI	56	43	44	98	60
GPT-4.1 MiniOpenAI	OpenAI	40	53	49	98	60
Nano Banana 2 (Gemini 3.1 Flash Image)Google	Google	50	98	48	43	60
GPT-5.4 Image 2OpenAI	OpenAI	50	98	48	43	60
GPT-5 Image MiniOpenAI	OpenAI	50	98	48	43	60
GPT-5 ImageOpenAI	OpenAI	50	98	48	43	60
GPT-4.1 NanoOpenAI	OpenAI	40	53	48	98	60
Sonar Pro SearchPerplexity	Perplexity	50	95	48	43	59
Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)Google	Google	50	93	48	43	59
GPT Chat LatestOpenAI	OpenAI	40	48	48	98	59
Nano Banana 2 (Gemini 3.1 Flash Image Preview)Google	Google	50	93	48	43	59
Nano Banana Pro (Gemini 3 Pro Image Preview)Google	Google	50	93	48	43	59
GPT-4OpenAI	OpenAI	53	40	42	98	58
GPT-4o-mini (2024-07-18)OpenAI	OpenAI	42	48	42	98	58
Gemini Omni 1.2 FlashGoogle	Google	45	98	48	35	57
Gemini Omni 1.1 FlashGoogle	Google	45	98	48	35	57
Ministral 3 14B 2512Mistral AI	Mistral AI	40	48	40	98	57
Ministral 3 8B 2512Mistral AI	Mistral AI	40	48	40	98	57
Ministral 3 3B 2512Mistral AI	Mistral AI	40	48	40	98	57
Qwen3 VL 8B InstructAlibaba	Alibaba	40	48	40	98	57
Qwen3 VL 30B A3B InstructAlibaba	Alibaba	40	48	40	98	57
Qwen3 VL 235B A22B InstructAlibaba	Alibaba	40	48	40	98	57
Mistral Medium 3.1Mistral AI	Mistral AI	40	48	40	98	57
Mistral Small 3.2 24BMistral AI	Mistral AI	40	48	40	98	57
Mistral Medium 3Mistral AI	Mistral AI	40	48	40	98	57
Mistral Small 3.1 24BMistral AI	Mistral AI	40	48	40	98	57
Gemma 3 12BGoogle	Google	40	48	40	98	57
Gemma 3 27BGoogle	Google	40	48	40	98	57
Sonar Reasoning ProPerplexity	Perplexity	45	98	48	35	57
Command ACohere	Cohere	46	45	36	98	56
Perceptron Mk1perceptron	perceptron	50	90	40	43	56
Pareto 26.9unbiased	unbiased	37	48	38	98	55
Qwen3 Coder FlashAlibaba	Alibaba	40	53	30	98	55
Hunyuan A13B InstructTencent	Tencent	50	98	30	43	55
Qwen-PlusAlibaba	Alibaba	40	53	30	98	55
Nemotron 3.5 Content SafetyNVIDIA	NVIDIA	45	98	40	35	55
Reka Edgerekaai	rekaai	40	40	40	98	55
Nova Premier 1.0Amazon	Amazon	35	53	40	90	55
Ripple 65B Previewricursiveai	ricursiveai	37	53	28	98	54
Qwen3 Coder NextAlibaba	Alibaba	40	48	30	98	54
GPT AudioOpenAI	OpenAI	40	48	30	98	54
GPT Audio MiniOpenAI	OpenAI	40	48	30	98	54
Devstral 2 2512Mistral AI	Mistral AI	40	48	30	98	54
Qwen3 Next 80B A3B InstructAlibaba	Alibaba	40	48	30	98	54
Kimi K2 0905Moonshot AI	Moonshot AI	40	48	30	98	54
Codestral 2508Mistral AI	Mistral AI	40	48	30	98	54
Qwen3 Coder 30B A3B InstructAlibaba	Alibaba	40	48	30	98	54
Qwen3 30B A3B Instruct 2507Alibaba	Alibaba	40	48	30	98	54
Qwen3 Coder 480B A35BAlibaba	Alibaba	40	48	30	98	54
Reka Flash 3rekaai	rekaai	50	93	30	43	54
Sonar Deep ResearchPerplexity	Perplexity	45	98	38	35	54
Mistral NemoMistral AI	Mistral AI	40	48	30	98	54
Pareto 26.10unbiased	unbiased	32	53	38	90	53
Pareto Latest~unbiased	~unbiased	32	53	38	90	53
Command R7B (12-2024)Cohere	Cohere	40	45	30	98	53
Llama 3.1 8B InstructMeta	Meta	38	48	28	98	53
Voxtral Small 24B 2507Mistral AI	Mistral AI	40	43	30	98	53
SabaMistral AI	Mistral AI	40	43	30	98	53
Qwen2.5 7B InstructAlibaba	Alibaba	40	43	30	98	53
Qwen2.5 72B InstructAlibaba	Alibaba	40	43	30	98	53
Nova Lite 1.0Amazon	Amazon	35	45	40	90	53
Nova Pro 1.0Amazon	Amazon	35	45	40	90	53
Command R (08-2024)Cohere	Cohere	37	45	28	98	52
Command R+ (08-2024)Cohere	Cohere	37	45	28	98	52
GPT-3.5 Turbo (older v0613)OpenAI	OpenAI	40	40	30	98	52
GPT-3.5 Turbo 16kOpenAI	OpenAI	40	40	30	98	52
Moondream 3 Previewmoondream	moondream	42	90	38	35	51
Moondream 3.1moondream	moondream	42	90	38	35	51
Gemma 2 27BGoogle	Google	67	40	54	43	51
Kimi K2 0711Moonshot AI	Moonshot AI	35	48	30	90	51
Nova Micro 1.0Amazon	Amazon	35	45	30	90	50
Phi 4Microsoft	Microsoft	54	40	43	43	45
Lyria 3 Pro PreviewGoogle	Google	40	53	40	43	44
Lyria 3 Clip PreviewGoogle	Google	40	53	40	43	44
Llama 3.1 70B InstructMeta	Meta	51	45	45	35	44
Nano Banana (Gemini 2.5 Flash Image)Google	Google	40	40	48	43	43
Gemma 3 4BGoogle	Google	40	48	40	43	43
Qwen2.5 VL 72B InstructAlibaba	Alibaba	40	48	40	43	43
UI-TARS 7B ByteDance	ByteDance	40	45	40	43	42
Sonar ProPerplexity	Perplexity	35	45	48	35	41
SonarPerplexity	Perplexity	35	43	48	35	40
Llama 3.2 3B InstructMeta	Meta	40	48	30	43	40
Extractscaledown	scaledown	32	53	38	35	40
Llama Guard 4 12BMeta	Meta	35	48	40	35	40
Mistral Small 3Mistral AI	Mistral AI	40	43	30	43	39
Qwen2.5 Coder 32B InstructAlibaba	Alibaba	40	43	30	43	39
Summarizescaledown	scaledown	32	50	38	35	39
Schematron V2 Turboinference-net	inference-net	37	45	28	43	38
Schematron V2 Smallinference-net	inference-net	37	45	28	43	38
Palmyra X5Writer	Writer	35	50	30	35	38
Hy-MT2-30B-A3BTencent	Tencent	37	40	28	43	37
Hy-MT2-7BTencent	Tencent	37	40	28	43	37
SwitchyardNVIDIA	NVIDIA	32	50	28	35	36
Llama 3.2 1B InstructMeta	Meta	35	43	30	35	36
autofixer-01Vercel	Vercel	35	40	30	35	35
MellumJetBrains	JetBrains	35	40	30	35	35
ALLaM 7B Instruct (preview)HUMAIN	HUMAIN	35	40	30	35	35
ALLaM 1 13B InstructHUMAIN	HUMAIN	35	40	30	35	35
ALLaM 2 7B InstructHUMAIN	HUMAIN	35	40	30	35	35
ALLaM 34BHUMAIN	HUMAIN	35	40	30	35	35
Falcon Mamba 7B InstructTII	TII	35	40	30	35	35
Hy-MT2-1.8BTencent	Tencent	32	40	28	35	34
Midjourney v6.1Midjourney	Midjourney	35	40	30	30	34
DALL-E 3OpenAI	OpenAI	35	40	30	30	34
Stable Diffusion 3.5Stability AI	Stability AI	35	40	30	30	34
FLUX.1 ProBlack Forest Labs	Black Forest Labs	35	40	30	30	34
Ideogram 2.0Ideogram	Ideogram	35	40	30	30	34
Recraft V3Recraft	Recraft	35	40	30	30	34
Imagen 3Google	Google	35	40	30	30	34
Adobe Firefly 3Adobe	Adobe	35	40	30	30	34
Leonardo PhoenixLeonardo AI	Leonardo AI	35	40	30	30	34
SoraOpenAI	OpenAI	35	40	30	30	34
Veo 2Google	Google	35	40	30	30	34
Runway Gen-3 AlphaRunway	Runway	35	40	30	30	34
Kling 1.6Kuaishou	Kuaishou	35	40	30	30	34
Pika 2.0Pika	Pika	35	40	30	30	34
MiniMax Video-01MiniMax	MiniMax	35	40	30	30	34
Luma Dream MachineLuma AI	Luma AI	35	40	30	30	34
Stable Video DiffusionStability AI	Stability AI	35	40	30	30	34
Wan 2.1 T2VWan AI	Wan AI	35	40	30	30	34
LTX-Video 2Lightricks	Lightricks	35	40	30	30	34

Explore More

Dive deeper with capability filters, pricing comparisons, or the full leaderboard.

All Explorers Capabilities Benchmarks Overview Leaderboard Compare Models

Frequently Asked Questions

AI benchmarks are standardized tests that measure model performance on specific tasks - math (MATH-500), coding (HumanEval, SWE-bench), reasoning (GPQA), knowledge (MMLU), and conversational ability (Arena ELO). They help compare models objectively.

No single benchmark tells the full story. Arena ELO (crowdsourced human preferences) best reflects real-world quality. SWE-bench is most relevant for coding. MATH-500 and GPQA test reasoning depth. Use multiple benchmarks for evaluation.

Benchmarks have limitations - models can be trained on benchmark data (contamination), and high scores don't always predict real-world performance. Arena ELO is hardest to game since it uses blind human comparisons. Use benchmarks as one signal among many.