Acropolis

GDPval

GDPval results that Epoch AI republishes. Despite the column label (%), the values are 0 to 1 fractions.

Upstream source
FinanceGeneral text
Independent. Run by an operator independent of the model vendors.Epoch AI Benchmarking Hub by Epoch AIVersion 1Data as of 17 Nov 2026

No filters. Showing 11 of 11 rows.

  1. 1
    OpenAIDossier

    none

    Win rate

    GPT-5.2, Win rate: 49.7% (scale 0.0 to 100.0)49.7%
    Win or tie rate
    70.9%
  2. 2
    AnthropicDossier

    Win rate

    Opus 4.5, Win rate: 45.5% (scale 0.0 to 100.0)45.5%
    Win or tie rate
    59.6%
  3. 3
    AnthropicDossier

    Win rate

    Opus 4.1, Win rate: 43.6% (scale 0.0 to 100.0)43.6%
    Win or tie rate
    47.6%
  4. 4
    AnthropicDossier

    Win rate

    Sonnet 4.5, Win rate: 42.5% (scale 0.0 to 100.0)42.5%
    Win or tie rate
    50.3%
  5. 5
    GoogleDossier

    Win rate

    Gemini 3 Pro Preview, Win rate: 40.3% (scale 0.0 to 100.0)40.3%
    Win or tie rate
    53.5%
  6. 6
    OpenAIDossier

    medium

    Win rate

    GPT-5, Win rate: 34.8% (scale 0.0 to 100.0)34.8%
    Win or tie rate
    38%
  7. 7
    OpenAIDossier

    medium

    Win rate

    o3, Win rate: 30.8% (scale 0.0 to 100.0)30.8%
    Win or tie rate
    34.1%
  8. 8
    OpenAIDossier

    high

    Win rate

    o4-mini, Win rate: 25.3% (scale 0.0 to 100.0)25.3%
    Win or tie rate
    27.8%
  9. 9
    GoogleDossier

    Win rate

    Gemini 2.5 Pro, Win rate: 23.3% (scale 0.0 to 100.0)23.3%
    Win or tie rate
    25.5%
  10. 10

    high

    Win rate

    Grok 4 (0709), Win rate: 21.1% (scale 0.0 to 100.0)21.1%
    Win or tie rate
    24.3%
  11. 11
    OpenAIDossier

    Win rate

    GPT-4o (Nov '24), Win rate: 9.9% (scale 0.0 to 100.0)9.9%
    Win or tie rate
    12.3%