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

CritPt

Independent testArtificial Analysis · updated Oct 8, 2026

What it measures

Seventy unpublished, research-level physics problems graded by an official server.

How to read it

One of the least saturated tests available — the best models solve under a third. A strong signal for original scientific reasoning.

Open at Artificial Analysis
Best
31.7%
Claude Opus 5.5
Median
15.4%
GLM-5.3-Flash

Results

39 of 41 tracked models have a result.

Filter in explorer
  1. 1
    Claude Opus 5.5

    Anthropic · Max effort

    31.7%
  2. 1
    GPT-6 Astra

    OpenAI · Max effort

    31.7%
  3. 1
    GPT-6.1 Sol

    OpenAI · Max effort

    31.7%
  4. 4
    Claude Sonnet 5.5

    Anthropic · Max effort

    31.4%
  5. 5
    GPT-5.6 Terra

    OpenAI · Max effort

    30.0%
  6. 6
    Claude Fable 5.1

    Anthropic · Max effort

    29.7%
  7. 7
    Gemini 4 ArgonLimited

    Google DeepMind · High effort

    27.1%
  8. 8
    MiMo-V2.6-ProOpen

    Xiaomi

    26.6%
  9. 9
    Muse Spark 1.3

    Meta · Max effort

    24.9%
  10. 10
    Kimi K3Open

    Moonshot AI · Max effort

    23.4%
  11. 11
    Step 5 Preview

    StepFun

    20.9%
  12. 12
    GPT-6 Luna

    OpenAI · Max effort

    19.4%
  13. 13
    GLM-5.3Open

    Z.ai · Max effort

    19.1%
  14. 14
    Claude Haiku 5.5

    Anthropic · Max effort

    18.9%
  15. 15
    Gemini 3.8 Flash

    Google DeepMind · High effort

    18.3%
  16. 16
    DeepSeek V4 ProOpen

    DeepSeek · Max effort

    18.0%
  17. 17
    Gemini 3.1 Pro

    Google DeepMind

    17.7%
  18. 17
    Grok 4.7

    SpaceXAI · xHigh effort

    17.7%
  19. 17
    Qwen3.8-Max

    Alibaba

    17.7%
  20. 20
    GLM-5.3-FlashOpen

    Z.ai

    15.4%
  21. 21
    DeepSeek V4.1 FlashOpen

    DeepSeek · Max effort

    14.3%
  22. 22
    MiMo-V2.6-FlashOpen

    Xiaomi

    12.0%
  23. 23
    Qwen3.8-FlashOpen

    Alibaba

    11.1%
  24. 24
    Mistral Large 4

    Mistral AI

    10.6%
  25. 25
    Kimi K2.7 CodeOpen

    Moonshot AI

    10.0%
  26. 26
    TM
    Inkling-SmallOpen

    Thinking Machines Lab

    8.3%
  27. 27
    TM
    InklingOpen

    Thinking Machines Lab · xHigh effort

    5.4%
  28. 27
    Qwen3.8-27BOpen

    Alibaba · xHigh effort

    5.4%
  29. 29
    Hy3Open

    Tencent

    4.9%
  30. 30
    MiniMax-M3Open

    MiniMax

    3.7%
  31. 31
    Nemotron 3 UltraOpen

    NVIDIA · Reasoning effort

    3.1%
  32. 32
    Muse Glimmer 30BOpen

    Meta · High effort

    2.6%
  33. 33
    Gemma 4 31BOpen

    Google DeepMind · Reasoning effort

    1.4%
  34. 34
    gpt-oss-120bOpen

    OpenAI · High effort

    1.1%
  35. 35
    Command A+Open

    Cohere

    0.3%
  36. 35
    Mistral Small 4Open

    Mistral AI · Reasoning effort

    0.3%
  37. 35
    Nova 2 Lite

    Amazon · High effort

    0.3%
  38. 38
    Gemini 3.5 Flash-Lite

    Google DeepMind

    0.0%
  39. 38
    Mistral Medium 3.5Open

    Mistral AI

    0.0%
2 tracked models without a result

Hy4 preview, MAI-Thinking-1.

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