Browse 420 LLMs

A long-horizon repository benchmark built to test realistic software engineering work. Its scores need a task-quality and setup check before they support a coding-agent decision.

1. 1
   
   Claude Mythos 580.3
2. 2
   
   Claude Fable 580.0
3. 3
   
   Claude Opus 579.2

63 models

A multilingual software-engineering benchmark for real-world code issue resolution across multiple programming languages.

1. 1
   
   Claude Opus 589.5
2. 2
   
   Claude Opus 4.884.4
3. 3
   Co
   Composer 2.579.8

36 models

An independently evaluated Tau3 banking benchmark from Artificial Analysis.

1. 1
   
   Grok 4.650.7
2. 2
   
   GLM-5.350.3
3. 3
   
   Qwen3.8-27B48.0

16 models

How important are skills for agents?

1. 1
   
   Grok 4.566.0
2. 2
   
   Gemini 3.7 Flash65.9
3. 3
   
   GPT-5.5 Codex62.6

29 models

A broad research-agent benchmark for open-ended information gathering, synthesis, and answer construction across wide search spaces.

1. 1
   Me
   Mercury 292.3
2. 2
   
   Qwen3.8 Max81.9
3. 3
   
   Kimi K2.680.8

14 models

A benchmark for web-browsing agents that must search, inspect sources, gather evidence, and return the correct answer to research-oriented questions.

1. 1
   
   GPT-5.6 Sol92.2
2. 2
   
   Kimi K391.2
3. 3
   
   Claude Opus 590.8

40 models

MRCR v2 slice focused on very long contexts at 128K-256K lengths.

1. 1
   
   GPT-5.587.5
2. 2
   
   Claude Opus 4.7 (Adaptive)59.2

2 models

Human-validated software engineering issues from real-world Python repositories.

1. 1
   
   Claude Opus 4.7 (max)83.5
2. 2
   
   GPT-5.5 (xhigh)80.6
3. 3
   
   Gemini 3.5 Flash (high)79.3

33 models

Factoid questions spanning politics, science, technology, art, sports, geography, and music.

1. 1
   
   Gemini 3.1 Pro Preview77.3
2. 2
   
   Gemini 3 Pro Preview72.9
3. 3
   
   GPT-5.6 Sol (max)71.6

78 models

Abstract reasoning and pattern generalization on grid-based tasks.

1. 1
   
   GPT-5.6 Sol (max)92.5
2. 2
   
   Claude Opus 5 (max)90.4
3. 3
   
   GPT-5.6 Sol (xhigh)90.0

184 models

Exceptionally difficult research-level mathematics problems.

1. 1
   
   Claude Fable 5 (max)87.8
2. 2
   
   GPT-5.6 Sol (max)82.9
3. 3
   
   GPT-5.6 Sol (pro, max)80.5

51 models

Real-world cybersecurity evaluation of AI agents reproducing vulnerabilities with working proof-of-concept tests.

1. 1
   
   DeepSeek-V4-Flash93.2
2. 2
   
   DeepSeek-V4-Flash91.2
3. 3
   
   Multi-model (GPT-5.4, Claude Opus 4.6, Claude Sonnet 4.6)91.0

51 models

Composite Artificial Analysis index across mathematics, science, coding, and reasoning evaluations.

1. 1
   
   Claude Opus 5 (max)63.1
2. 2
   
   Claude Opus 5 (xhigh)62.5
3. 3
   
   Claude Fable 5 (with fallback)62.1

603 models

Artificial Analysis composite index for programming and software-engineering capability.

1. 1
   
   GPT-5.6 Sol (xhigh)78.3
2. 2
   
   Claude Opus 5 (max)78.0
3. 3
   
   GPT-5.6 Sol (max)77.4

228 models

Artificial Analysis composite index for tool use, planning, autonomy, and complex agentic workflows.

1. 1
   
   Claude Opus 5 (max)59.2
2. 2
   
   GLM-5.3 (max)59.1
3. 3
   
   Grok 4.6 (high)58.7

170 models

Artificial Analysis factual-knowledge and hallucination-resistance evaluation.

1. 1
   
   Claude Fable 5 (with fallback)43.3
2. 2
   
   Claude Opus 5 (max)37.1
3. 3
   
   Claude Opus 5 (xhigh)35.4

81 models

Graduate-level, expert-written science questions evaluated by Artificial Analysis.

1. 1
   
   Grok 4.6 (high)94.9
2. 2
   
   Gemini 3.7 Flash (high)94.5
3. 3
   
   Gemini 3.1 Pro Preview94.1

584 models

Broad expert-level reasoning and knowledge evaluation published by Artificial Analysis.

1. 1
   
   Claude Fable 5 (with fallback)55.5
2. 2
   
   Claude Opus 5 (max)54.9
3. 3
   
   Claude Opus 5 (xhigh)54.4

575 models

Instruction-following benchmark results published by Artificial Analysis.

1. 1
   
   Grok 4.3 (medium)83.3
2. 2
   
   Grok 4.20 030982.9
3. 3
   
   MiniMax-M382.9

450 models

Scientific programming benchmark results published by Artificial Analysis.

1. 1
   
   Claude Fable 5 (with fallback)60.2
2. 2
   
   Gemini 3.1 Pro Preview58.9
3. 3
   
   Kimi K3 (max)58.7

575 models

Agentic terminal and software-engineering results published by Artificial Analysis.

1. 1
   
   GPT-5.6 Sol (xhigh)89.5
2. 2
   
   Claude Opus 5 (max)89.1
3. 3
   
   Grok 4.6 (high)88.4

208 models

Critical-points programming evaluation published by Artificial Analysis.

1. 1
   
   GPT-5.6 Sol (max)32.3
2. 2
   
   GPT-5.5 Pro (xhigh)30.6
3. 3
   
   GPT-5.4 Pro (xhigh)30.0

490 models

Long-context reasoning results published by Artificial Analysis.

1. 1
   
   Muse Spark 1.2 (xhigh)83.3
2. 2
   
   Kimi K3 (max)82.7
3. 3
   
   Muse Spark 1.1 (xhigh)81.3

508 models

Tool-agent interaction benchmark results published by Artificial Analysis.

1. 1
   JT
   JT-35B-Flash99.1
2. 2
   
   GLM-5.2 (max)99.1
3. 3
   
   GLM-4.7-Flash98.8

440 models