Skip to content
Model Index
  • Overview
  • Models
  • Compare
  • Benchmarks
  • Finder
LabsMethodology
  1. Benchmarks
  2. Agents & knowledge work

GDPval-AA

Independent testArtificial Analysis · updated Oct 8, 2026

What it measures

Real deliverables — documents, spreadsheets, presentations — across 44 occupations from OpenAI's GDPval set, judged pairwise by a panel and converted to a normalised Elo scale (v2.1).

How to read it

Shown as a 0–100 normalised score. Higher means the model's work product wins more head-to-head comparisons.

Open at Artificial Analysis
Best
68.3
Claude Opus 5.5
Median
47.7
GPT-5.6 Terra

Results

39 of 41 tracked models have a result.

Filter in explorer
  1. 1
    Claude Opus 5.5

    Anthropic · Max effort

    68.3
  2. 2
    Claude Sonnet 5.5

    Anthropic · Max effort

    67.0
  3. 3
    Claude Fable 5.1

    Anthropic · Max effort

    62.9
  4. 4
    Grok 4.7

    SpaceXAI · xHigh effort

    60.8
  5. 5
    MiMo-V2.6-ProOpen

    Xiaomi

    59.3
  6. 6
    Muse Spark 1.3

    Meta · Max effort

    59.2
  7. 7
    Qwen3.8-Max

    Alibaba

    58.6
  8. 8
    GLM-5.3Open

    Z.ai · Max effort

    57.6
  9. 9
    GLM-5.3-FlashOpen

    Z.ai

    57.3
  10. 10
    Qwen3.8-FlashOpen

    Alibaba

    56.6
  11. 11
    Gemini 4 ArgonLimited

    Google DeepMind · High effort

    56.3
  12. 12
    Claude Haiku 5.5

    Anthropic · Max effort

    56.0
  13. 13
    MiMo-V2.6-FlashOpen

    Xiaomi

    55.5
  14. 14
    DeepSeek V4.1 FlashOpen

    DeepSeek · Max effort

    55.0
  15. 15
    Step 5 Preview

    StepFun

    54.2
  16. 16
    GPT-6.1 Sol

    OpenAI · Max effort

    53.8
  17. 17
    GPT-6 Astra

    OpenAI · Max effort

    52.1
  18. 18
    Kimi K3Open

    Moonshot AI · Max effort

    51.8
  19. 19
    DeepSeek V4 ProOpen

    DeepSeek · Max effort

    47.7
  20. 19
    GPT-5.6 Terra

    OpenAI · Max effort

    47.7
  21. 21
    GPT-6 Luna

    OpenAI · Max effort

    46.9
  22. 22
    Gemini 3.8 Flash

    Google DeepMind · High effort

    46.8
  23. 23
    Mistral Large 4

    Mistral AI

    46.2
  24. 23
    Qwen3.8-27BOpen

    Alibaba · xHigh effort

    46.2
  25. 25
    MiniMax-M3Open

    MiniMax

    37.3
  26. 26
    TM
    Inkling-SmallOpen

    Thinking Machines Lab

    31.2
  27. 27
    TM
    InklingOpen

    Thinking Machines Lab · xHigh effort

    28.9
  28. 28
    Hy3Open

    Tencent

    28.3
  29. 29
    Kimi K2.7 CodeOpen

    Moonshot AI

    27.0
  30. 30
    Nemotron 3 UltraOpen

    NVIDIA · Reasoning effort

    25.8
  31. 31
    Gemini 3.5 Flash-Lite

    Google DeepMind

    24.3
  32. 32
    Gemini 3.1 Pro

    Google DeepMind

    14.7
  33. 33
    Muse Glimmer 30BOpen

    Meta · High effort

    14.5
  34. 34
    Mistral Medium 3.5Open

    Mistral AI

    13.2
  35. 35
    Gemma 4 31BOpen

    Google DeepMind · Reasoning effort

    6.1
  36. 36
    gpt-oss-120bOpen

    OpenAI · High effort

    5.6
  37. 37
    Command A+Open

    Cohere

    0.9
  38. 38
    Mistral Small 4Open

    Mistral AI · Reasoning effort

    0.0
  39. 38
    Nova 2 Lite

    Amazon · High effort

    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