Acropolis

Legal Agent Benchmark (HLAB-Vals)

Legal Agent Benchmark (HLAB-Vals), a component of the Vals Index.

Upstream source
LegalAgentic
Independent. Run by an operator independent of the model vendors.Vals AI benchmarks by Vals AIVersion 1.1Data as of 17 Nov 2026

No filters. Showing 38 of 38 rows.

  1. 1
    GoogleDossier

    high

    Accuracy

    Gemini 4 Argon, Accuracy: 27.1% (scale 0.0 to 100.0), 95% confidence interval 19.8 to 34.427.1%(19.8–34.4)
    Duration per test
    1,755.4 s
    Cost per test
    $10.55
  2. 2

    max

    Tied #1

    Accuracy

    Muse Ember, Accuracy: 25% (scale 0.0 to 100.0), 95% confidence interval 17.8 to 32.225%(17.8–32.2)
    Duration per test
    353.6 s
    Cost per test
    $0.12
  3. 2

    xhigh

    Tied #1

    Accuracy

    Muse Spark Open, Accuracy: 25% (scale 0.0 to 100.0), 95% confidence interval 17.9 to 32.125%(17.9–32.1)
    Duration per test
    14,355.7 s
    Cost per test
    $13.63
  4. 4

    max

    Tied #1

    Accuracy

    Muse Wonder 1.0, Accuracy: 23.8% (scale 0.0 to 100.0), 95% confidence interval 16.6 to 30.923.8%(16.6–30.9)
    Duration per test
    906.1 s
    Cost per test
    $7.08
  5. 5

    max

    Tied #1

    Accuracy

    Muse Wonder Full 1.0, Accuracy: 23.3% (scale 0.0 to 100.0), 95% confidence interval 16.4 to 30.323.3%(16.4–30.3)
    Duration per test
    2,649.4 s
    Cost per test
    $5.41
  6. 6
    Thinking MachinesDossier
    Tied #1

    Accuracy

    Inkling 2 Small, Accuracy: 21.7% (scale 0.0 to 100.0), 95% confidence interval 15.1 to 28.221.7%(15.1–28.2)
    Duration per test
    1,290.9 s
    Cost per test
    $1.23
  7. 7
    AlibabaDossier

    xhigh

    Tied #1

    Accuracy

    Qwen4 Max, Accuracy: 21.3% (scale 0.0 to 100.0), 95% confidence interval 14.7 to 27.821.3%(14.7–27.8)
    Duration per test
    4,454.1 s
    Cost per test
    $12.87
  8. 8

    xhigh

    Tied #1

    Accuracy

    Grok 4.7, Accuracy: 18.3% (scale 0.0 to 100.0), 95% confidence interval 12.1 to 24.618.3%(12.1–24.6)
    Duration per test
    2,148.4 s
    Cost per test
    $10.75
  9. 9
    GoogleDossier

    high

    Accuracy

    Gemini 4 Carbon, Accuracy: 17.5% (scale 0.0 to 100.0), 95% confidence interval 11.6 to 23.417.5%(11.6–23.4)
    Duration per test
    1,720.8 s
    Cost per test
    $6.52
  10. 10
    MistralDossier

    high

    Accuracy

    Mistral Large 4 OSS, Accuracy: 15.8% (scale 0.0 to 100.0), 95% confidence interval 9.7 to 22.015.8%(9.7–22.0)
    Duration per test
    688.7 s
    Cost per test
    $0.73
  11. 11
    AlibabaDossier

    Accuracy

    Qwen3.8 Max Internal, Accuracy: 15.4% (scale 0.0 to 100.0), 95% confidence interval 9.4 to 21.415.4%(9.4–21.4)
    Duration per test
    2,950.6 s
    Cost per test
    $3.36
  12. 12

    max

    Accuracy

    GLM-5.3, Accuracy: 14.6% (scale 0.0 to 100.0), 95% confidence interval 9.0 to 20.214.6%(9.0–20.2)
    Duration per test
    1,911 s
    Cost per test
    $3.99
  13. 13
    DeepSeekDossier

    max

    Accuracy

    DeepSeek V4.1 Flash 1017, Accuracy: 14.2% (scale 0.0 to 100.0), 95% confidence interval 8.7 to 19.614.2%(8.7–19.6)
    Duration per test
    1,328.1 s
    Cost per test
    $0.61
  14. 14
    XiaomiDossier

    Accuracy

    MiMo-V2.6-Pro, Accuracy: 13.8% (scale 0.0 to 100.0), 95% confidence interval 8.2 to 19.313.8%(8.2–19.3)
    Duration per test
    1,861.8 s
    Cost per test
    $0.22
  15. 15
    XiaomiDossier

    Accuracy

    MiMo-V2.6-Flash, Accuracy: 13.3% (scale 0.0 to 100.0), 95% confidence interval 7.8 to 18.913.3%(7.8–18.9)
    Duration per test
    1,342.6 s
    Cost per test
    $0.086
  16. 16
    Moonshot AIDossier

    max

    Accuracy

    Kimi K3, Accuracy: 12.9% (scale 0.0 to 100.0), 95% confidence interval 7.4 to 18.412.9%(7.4–18.4)
    Duration per test
    2,556.3 s
    Cost per test
    $4.14
  17. 16
    StepFunDossier

    Accuracy

    Step 5 Preview, Accuracy: 12.9% (scale 0.0 to 100.0), 95% confidence interval 7.4 to 18.412.9%(7.4–18.4)
    Duration per test
    993.2 s
    Cost per test
    $1.13
  18. 18
    TencentDossier

    Accuracy

    Hy4, Accuracy: 12.5% (scale 0.0 to 100.0), 95% confidence interval 7.3 to 17.712.5%(7.3–17.7)
    Duration per test
    1,167.2 s
    Cost per test
    $1.04
  19. 19
    GoogleDossier

    high

    Accuracy

    Gemini 3.8 Flash, Accuracy: 12.1% (scale 0.0 to 100.0), 95% confidence interval 7.0 to 17.112.1%(7.0–17.1)
    Duration per test
    1,161.7 s
    Cost per test
    $3.66
  20. 20
    Moonshot AIDossier

    max

    Accuracy

    Kimi K3.1 Pro, Accuracy: 11.7% (scale 0.0 to 100.0), 95% confidence interval 6.5 to 16.811.7%(6.5–16.8)
    Duration per test
    4,373.4 s
    Cost per test
    $9.00
  21. 21
    GoogleDossier

    xhigh

    Accuracy

    Gemini 4 Hydrogen, Accuracy: 11.3% (scale 0.0 to 100.0), 95% confidence interval 6.0 to 16.511.3%(6.0–16.5)
    Duration per test
    742.8 s
    Cost per test
    $3.28
  22. 22
    DeepSeekDossier

    max

    Accuracy

    DeepSeek V4 Pro 0813, Accuracy: 10.8% (scale 0.0 to 100.0), 95% confidence interval 5.5 to 16.210.8%(5.5–16.2)
    Duration per test
    655.4 s
    Cost per test
    $0.80
  23. 23

    max

    Accuracy

    GLM-5.3 Flash, Accuracy: 10.4% (scale 0.0 to 100.0), 95% confidence interval 5.4 to 15.510.4%(5.4–15.5)
    Duration per test
    744.8 s
    Cost per test
    $0.32
  24. 23
    AnthropicDossier

    max

    Accuracy

    Sonnet 5.5, Accuracy: 10.4% (scale 0.0 to 100.0), 95% confidence interval 5.4 to 15.510.4%(5.4–15.5)
    Duration per test
    4,398 s
    Cost per test
    $17.71
  25. 25
    AnthropicDossier

    max

    Accuracy

    Opus 5.5, Accuracy: 9.2% (scale 0.0 to 100.0), 95% confidence interval 4.5 to 13.89.2%(4.5–13.8)
    Duration per test
    3,507.4 s
    Cost per test
    $21.81
  26. 26
    TencentDossier

    Accuracy

    Hy4 Preview, Accuracy: 8.8% (scale 0.0 to 100.0), 95% confidence interval 4.5 to 13.08.8%(4.5–13.0)
    Duration per test
    1,989.3 s
    Cost per test
    $0.98
  27. 27
    MiniMaxDossier

    max

    Accuracy

    MiniMax M3.1 Flash, Accuracy: 7.5% (scale 0.0 to 100.0), 95% confidence interval 3.1 to 11.97.5%(3.1–11.9)
    Duration per test
    1,497.6 s
    Cost per test
    $0.47
  28. 28
    OpenAIDossier

    max

    Accuracy

    GPT-6 Astra, Accuracy: 7.1% (scale 0.0 to 100.0), 95% confidence interval 3.4 to 10.87.1%(3.4–10.8)
    Duration per test
    1,953.4 s
    Cost per test
    $25.46
  29. 29
    OpenAIDossier

    max

    Accuracy

    GPT-6.1 Sol, Accuracy: 6.3% (scale 0.0 to 100.0), 95% confidence interval 2.7 to 9.86.3%(2.7–9.8)
    Duration per test
    2,425.6 s
    Cost per test
    $3.54
  30. 30
    AnthropicDossier

    high

    Accuracy

    Fable 5.2, Accuracy: 5% (scale 0.0 to 100.0), 95% confidence interval 1.6 to 8.45%(1.6–8.4)
    Duration per test
    1,954.2 s
    Cost per test
    $19.37
  31. 30
    InceptionDossier

    high

    Accuracy

    Mercury 3, Accuracy: 5% (scale 0.0 to 100.0), 95% confidence interval 1.8 to 8.25%(1.8–8.2)
    Duration per test
    886.9 s
    Cost per test
    $0.19
  32. 32
    ReflectionDossier

    Accuracy

    Beam 1.0, Accuracy: 4.6% (scale 0.0 to 100.0), 95% confidence interval 1.1 to 8.14.6%(1.1–8.1)
    Duration per test
    116.9 s
    Cost per test
    Not tested
  33. 33
    OpenAIDossier

    max

    Accuracy

    GPT-6 Luna, Accuracy: 4.2% (scale 0.0 to 100.0), 95% confidence interval 1.0 to 7.44.2%(1.0–7.4)
    Duration per test
    1,305.5 s
    Cost per test
    $0.30
  34. 33
    AnthropicDossier

    max

    Accuracy

    Haiku 5.5, Accuracy: 4.2% (scale 0.0 to 100.0), 95% confidence interval 1.2 to 7.14.2%(1.2–7.1)
    Duration per test
    2,620.8 s
    Cost per test
    $4.26
  35. 35
    Thinking MachinesDossier

    Accuracy

    Inkling, Accuracy: 3.8% (scale 0.0 to 100.0), 95% confidence interval 0.9 to 6.63.8%(0.9–6.6)
    Duration per test
    529.6 s
    Cost per test
    $1.25
  36. 35
    NVIDIADossier

    high

    Accuracy

    Nemotron 3.5 Super VL, Accuracy: 3.8% (scale 0.0 to 100.0), 95% confidence interval 1.1 to 6.43.8%(1.1–6.4)
    Duration per test
    172.1 s
    Cost per test
    $0.37
  37. 37
    Thinking MachinesDossier

    Accuracy

    Inkling Small, Accuracy: 2.9% (scale 0.0 to 100.0), 95% confidence interval 0.2 to 5.62.9%(0.2–5.6)
    Duration per test
    455.3 s
    Cost per test
    $0.18
  38. 38
    OpenAIDossier

    max

    Accuracy

    GPT-6.1 Terra, Accuracy: 1.7% (scale 0.0 to 100.0), 95% confidence interval 0.5 to 5.41.7%(0.5–5.4)
    Duration per test
    821.5 s
    Cost per test
    $1.48