Acropolis

Legal Research

Legal Research, 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 2.2Data as of 17 Nov 2026

No filters. Showing 38 of 38 rows.

  1. 1
    GoogleDossier

    high

    Accuracy

    Gemini 4 Carbon, Accuracy: 55.8% (scale 0.0 to 100.0), 95% confidence interval 49.0 to 62.555.8%(49.0–62.5)
    Duration per test
    3,112.6 s
    Cost per test
    $8.74
  2. 2
    GoogleDossier

    high

    Tied #1

    Accuracy

    Gemini 4 Argon, Accuracy: 54.8% (scale 0.0 to 100.0), 95% confidence interval 48.0 to 61.654.8%(48.0–61.6)
    Duration per test
    1,881 s
    Cost per test
    $6.45
  3. 3
    GoogleDossier

    xhigh

    Tied #1

    Accuracy

    Gemini 4 Hydrogen, Accuracy: 53.8% (scale 0.0 to 100.0), 95% confidence interval 47.1 to 60.653.8%(47.1–60.6)
    Duration per test
    1,964.3 s
    Cost per test
    $6.27
  4. 4

    max

    Tied #1

    Accuracy

    Muse Wonder Full 1.0, Accuracy: 53.4% (scale 0.0 to 100.0), 95% confidence interval 46.6 to 60.253.4%(46.6–60.2)
    Duration per test
    1,654.4 s
    Cost per test
    $3.60
  5. 5

    max

    Tied #1

    Accuracy

    Muse Wonder 1.0, Accuracy: 50.5% (scale 0.0 to 100.0), 95% confidence interval 43.7 to 57.350.5%(43.7–57.3)
    Duration per test
    927.1 s
    Cost per test
    $3.98
  6. 5
    AnthropicDossier

    max

    Tied #1

    Accuracy

    Opus 5.5, Accuracy: 50.5% (scale 0.0 to 100.0), 95% confidence interval 43.7 to 57.350.5%(43.7–57.3)
    Duration per test
    6,749.5 s
    Cost per test
    $23.61
  7. 7

    max

    Tied #1

    Accuracy

    GLM-5.3, Accuracy: 49% (scale 0.0 to 100.0), 95% confidence interval 42.2 to 55.849%(42.2–55.8)
    Duration per test
    3,110.2 s
    Cost per test
    $2.24
  8. 7

    max

    Tied #1

    Accuracy

    Muse Ember, Accuracy: 49% (scale 0.0 to 100.0), 95% confidence interval 42.2 to 55.849%(42.2–55.8)
    Duration per test
    207.7 s
    Cost per test
    $0.057
  9. 9
    AnthropicDossier

    max

    Tied #1

    Accuracy

    Sonnet 5.5, Accuracy: 48.1% (scale 0.0 to 100.0), 95% confidence interval 41.3 to 54.948.1%(41.3–54.9)
    Duration per test
    5,589 s
    Cost per test
    $13.66
  10. 10

    xhigh

    Tied #1

    Accuracy

    Grok 4.7, Accuracy: 47.1% (scale 0.0 to 100.0), 95% confidence interval 40.3 to 53.947.1%(40.3–53.9)
    Duration per test
    1,274.3 s
    Cost per test
    $4.92
  11. 10
    XiaomiDossier
    Tied #1

    Accuracy

    MiMo-V2.6-Pro, Accuracy: 47.1% (scale 0.0 to 100.0), 95% confidence interval 40.3 to 53.947.1%(40.3–53.9)
    Duration per test
    1,816.3 s
    Cost per test
    $0.18
  12. 12
    Thinking MachinesDossier
    Tied #1

    Accuracy

    Inkling 2 Small, Accuracy: 46.6% (scale 0.0 to 100.0), 95% confidence interval 39.8 to 53.446.6%(39.8–53.4)
    Duration per test
    735.2 s
    Cost per test
    $0.29
  13. 13
    AnthropicDossier

    high

    Accuracy

    Fable 5.2, Accuracy: 46.2% (scale 0.0 to 100.0), 95% confidence interval 39.4 to 52.946.2%(39.4–52.9)
    Duration per test
    416.2 s
    Cost per test
    $3.96
  14. 13
    Moonshot AIDossier

    max

    Accuracy

    Kimi K3, Accuracy: 46.2% (scale 0.0 to 100.0), 95% confidence interval 39.4 to 52.946.2%(39.4–52.9)
    Duration per test
    760.1 s
    Cost per test
    $3.47
  15. 13
    AlibabaDossier

    Accuracy

    Qwen3.8 Max Internal, Accuracy: 46.2% (scale 0.0 to 100.0), 95% confidence interval 39.4 to 52.946.2%(39.4–52.9)
    Duration per test
    7,649.6 s
    Cost per test
    $5.31
  16. 16
    DeepSeekDossier

    max

    Accuracy

    DeepSeek V4.1 Flash 1017, Accuracy: 45.2% (scale 0.0 to 100.0), 95% confidence interval 38.4 to 52.045.2%(38.4–52.0)
    Duration per test
    2,106.9 s
    Cost per test
    $0.70
  17. 16

    max

    Accuracy

    GLM-5.3 Flash, Accuracy: 45.2% (scale 0.0 to 100.0), 95% confidence interval 38.4 to 52.045.2%(38.4–52.0)
    Duration per test
    3,067.5 s
    Cost per test
    $0.24
  18. 16
    TencentDossier

    Accuracy

    Hy4 Preview, Accuracy: 45.2% (scale 0.0 to 100.0), 95% confidence interval 38.4 to 52.045.2%(38.4–52.0)
    Duration per test
    3,839.5 s
    Cost per test
    $0.85
  19. 19
    AnthropicDossier

    max

    Accuracy

    Haiku 5.5, Accuracy: 43.8% (scale 0.0 to 100.0), 95% confidence interval 37.0 to 50.543.8%(37.0–50.5)
    Duration per test
    4,979.8 s
    Cost per test
    $1.65
  20. 20
    TencentDossier

    Accuracy

    Hy4, Accuracy: 41.3% (scale 0.0 to 100.0), 95% confidence interval 34.6 to 48.141.3%(34.6–48.1)
    Duration per test
    5,455 s
    Cost per test
    $2.22
  21. 20
    AlibabaDossier

    xhigh

    Accuracy

    Qwen4 Max, Accuracy: 41.3% (scale 0.0 to 100.0), 95% confidence interval 34.6 to 48.141.3%(34.6–48.1)
    Duration per test
    3,438.2 s
    Cost per test
    $2.59
  22. 22
    DeepSeekDossier

    max

    Accuracy

    DeepSeek V4 Pro 0813, Accuracy: 40.9% (scale 0.0 to 100.0), 95% confidence interval 34.2 to 47.640.9%(34.2–47.6)
    Duration per test
    2,527.2 s
    Cost per test
    $1.09
  23. 23
    StepFunDossier

    Accuracy

    Step 5 Preview, Accuracy: 40.4% (scale 0.0 to 100.0), 95% confidence interval 33.7 to 47.140.4%(33.7–47.1)
    Duration per test
    2,168.4 s
    Cost per test
    $0.84
  24. 24
    OpenAIDossier

    max

    Accuracy

    GPT-6 Astra, Accuracy: 39.4% (scale 0.0 to 100.0), 95% confidence interval 32.8 to 46.139.4%(32.8–46.1)
    Duration per test
    1,491.3 s
    Cost per test
    $10.58
  25. 25
    GoogleDossier

    high

    Accuracy

    Gemini 3.8 Flash, Accuracy: 38.9% (scale 0.0 to 100.0), 95% confidence interval 32.3 to 45.638.9%(32.3–45.6)
    Duration per test
    324.3 s
    Cost per test
    $1.63
  26. 26
    OpenAIDossier

    max

    Accuracy

    GPT-6.1 Sol, Accuracy: 38.5% (scale 0.0 to 100.0), 95% confidence interval 31.8 to 45.138.5%(31.8–45.1)
    Duration per test
    2,060.5 s
    Cost per test
    $3.38
  27. 26

    xhigh

    Accuracy

    Muse Spark Open, Accuracy: 38.5% (scale 0.0 to 100.0), 95% confidence interval 31.8 to 45.138.5%(31.8–45.1)
    Duration per test
    1,763 s
    Cost per test
    $1.56
  28. 28
    XiaomiDossier

    Accuracy

    MiMo-V2.6-Flash, Accuracy: 38% (scale 0.0 to 100.0), 95% confidence interval 31.4 to 44.638%(31.4–44.6)
    Duration per test
    919.7 s
    Cost per test
    $0.070
  29. 29
    Moonshot AIDossier

    max

    Accuracy

    Kimi K3.1 Pro, Accuracy: 37.5% (scale 0.0 to 100.0), 95% confidence interval 30.9 to 44.137.5%(30.9–44.1)
    Duration per test
    6,237.4 s
    Cost per test
    $7.23
  30. 30
    MistralDossier

    high

    Accuracy

    Mistral Large 4 OSS, Accuracy: 35.6% (scale 0.0 to 100.0), 95% confidence interval 29.1 to 42.135.6%(29.1–42.1)
    Duration per test
    1,704.1 s
    Cost per test
    $1.09
  31. 31
    MiniMaxDossier

    max

    Accuracy

    MiniMax M3.1 Flash, Accuracy: 34.1% (scale 0.0 to 100.0), 95% confidence interval 27.7 to 40.634.1%(27.7–40.6)
    Duration per test
    2,912 s
    Cost per test
    $0.44
  32. 32
    OpenAIDossier

    max

    Accuracy

    GPT-6 Luna, Accuracy: 30.3% (scale 0.0 to 100.0), 95% confidence interval 24.0 to 36.530.3%(24.0–36.5)
    Duration per test
    2,547.7 s
    Cost per test
    $0.44
  33. 33
    ReflectionDossier

    Accuracy

    Beam 1.0, Accuracy: 28.8% (scale 0.0 to 100.0), 95% confidence interval 22.7 to 35.028.8%(22.7–35.0)
    Duration per test
    94.6 s
    Cost per test
    Not tested
  34. 33
    InceptionDossier

    high

    Accuracy

    Mercury 3, Accuracy: 28.8% (scale 0.0 to 100.0), 95% confidence interval 22.7 to 35.028.8%(22.7–35.0)
    Duration per test
    446.5 s
    Cost per test
    $0.074
  35. 35
    Thinking MachinesDossier

    Accuracy

    Inkling, Accuracy: 28.4% (scale 0.0 to 100.0), 95% confidence interval 22.2 to 34.528.4%(22.2–34.5)
    Duration per test
    869.4 s
    Cost per test
    $0.49
  36. 36
    OpenAIDossier

    max

    Accuracy

    GPT-6.1 Terra, Accuracy: 26% (scale 0.0 to 100.0), 95% confidence interval 20.0 to 31.926%(20.0–31.9)
    Duration per test
    4,202.7 s
    Cost per test
    $3.22
  37. 37
    Thinking MachinesDossier

    Accuracy

    Inkling Small, Accuracy: 25.5% (scale 0.0 to 100.0), 95% confidence interval 19.5 to 31.425.5%(19.5–31.4)
    Duration per test
    774.8 s
    Cost per test
    $0.14
  38. 38
    NVIDIADossier

    high

    Accuracy

    Nemotron 3.5 Super VL, Accuracy: 10.1% (scale 0.0 to 100.0), 95% confidence interval 6.0 to 14.210.1%(6.0–14.2)
    Duration per test
    83 s
    Cost per test
    $0.064