Skip to content
Model Index
  • Overview
  • Models
  • Compare
  • Benchmarks
  • Finder
LabsMethodology
  1. Benchmarks
  2. Reasoning & science

GPQA Diamond

Independent testSaturatingArtificial Analysis · updated Oct 8, 2026

What it measures

Graduate-level multiple-choice questions in physics, chemistry and biology that resist simple web search.

How to read it

Saturated: the leaders sit in the mid-90s and Artificial Analysis has retired it to its legacy set, so the newest models are often not run. Kept for continuity with older comparisons.

Open at Artificial Analysis
Best
96.1%
GPT-6 Astra
Median
90.5%
Qwen3.8-27B

Results

27 of 41 tracked models have a result.

Filter in explorer
  1. 1
    GPT-6 Astra

    OpenAI · Max effort

    96.1%
  2. 2
    Gemini 3.8 Flash

    Google DeepMind · High effort

    95.3%
  3. 3
    Gemini 3.1 Pro

    Google DeepMind

    94.1%
  4. 4
    Claude Fable 5.1

    Anthropic · Max effort

    93.7%
  5. 5
    Kimi K3Open

    Moonshot AI · Max effort

    93.5%
  6. 5
    Muse Spark 1.3

    Meta · Max effort

    93.5%
  7. 7
    MiniMax-M3Open

    MiniMax

    92.9%
  8. 8
    DeepSeek V4 ProOpen

    DeepSeek · Max effort

    92.8%
  9. 8
    Qwen3.8-Max

    Alibaba

    92.8%
  10. 10
    GPT-5.6 Terra

    OpenAI · Max effort

    92.5%
  11. 11
    Qwen3.8-FlashOpen

    Alibaba

    92.3%
  12. 12
    GLM-5.3Open

    Z.ai · Max effort

    91.7%
  13. 13
    GLM-5.3-FlashOpen

    Z.ai

    91.2%
  14. 14
    Qwen3.8-27BOpen

    Alibaba · xHigh effort

    90.5%
  15. 15
    Hy3Open

    Tencent

    89.7%
  16. 16
    Kimi K2.7 CodeOpen

    Moonshot AI

    89.6%
  17. 17
    TM
    Inkling-SmallOpen

    Thinking Machines Lab

    89.5%
  18. 18
    TM
    InklingOpen

    Thinking Machines Lab · xHigh effort

    87.2%
  19. 19
    Nemotron 3 UltraOpen

    NVIDIA · Reasoning effort

    86.7%
  20. 20
    Gemma 4 31BOpen

    Google DeepMind · Reasoning effort

    85.7%
  21. 21
    Gemini 3.5 Flash-Lite

    Google DeepMind

    83.8%
  22. 22
    Muse Glimmer 30BOpen

    Meta · High effort

    83.5%
  23. 23
    Nova 2 Lite

    Amazon · High effort

    81.1%
  24. 24
    gpt-oss-120bOpen

    OpenAI · High effort

    78.2%
  25. 25
    Mistral Small 4Open

    Mistral AI · Reasoning effort

    76.9%
  26. 26
    Command A+Open

    Cohere

    76.1%
  27. 27
    Mistral Medium 3.5Open

    Mistral AI

    74.8%
14 tracked models without a result

Claude Haiku 5.5, Claude Opus 5.5, Claude Sonnet 5.5, DeepSeek V4.1 Flash, Gemini 4 Argon, GPT-6 Luna, GPT-6.1 Sol, Grok 4.7, Hy4 preview, MAI-Thinking-1, MiMo-V2.6-Flash, MiMo-V2.6-Pro, Mistral Large 4, Step 5 Preview.

Missing means the evaluator has not published a run, not that the model would score poorly.

Model Index

An independent snapshot of frontier SI models and how they perform on third-party benchmarks.

Data as of Oct 8, 2026

Explore

  • All models
  • Compare
  • Benchmarks
  • Model finder
  • Labs
  • Methodology & sources

Sources

Benchmark results come from Artificial Analysis, Arena, Vals AI and ARC Prize. Specifications and prices come from each lab's own documentation.

Not affiliated with any lab or evaluator. Names and logos are trademarks of their respective owners and are used only to identify them. Scores change often — check the linked source before making a decision that depends on one.

  • Overview
  • Models
  • Compare
  • Benchmarks
  • Finder
  • Overview
  • Models
  • Compare
  • Benchmarks
  • Finder
  • Overview
  • Models
  • Compare
  • Benchmarks
  • Finder