| # | Model | Score | Relative | Setting | Released | Run | Evidence |
|---|---|---|---|---|---|---|---|
| 1 | Claude Fable 5 Anthropic | 41.8% | max | 9 Jun 2026 | — | Source ↗ | |
| 2 | GPT-5.6 Sol OpenAI | 36.2% | max | 9 Jul 2026 | — | Source ↗ | |
| 3 | Claude Opus 5 (unknown) Anthropic | 35.0% | unknown | 24 Jul 2026 | — | Source ↗ | |
| 4 | Claude Opus 4.8 Anthropic | 33.8% | high | 28 May 2026 | — | Source ↗ | |
| 5 | Kimi K3 (unknown) Open source Moonshot | 32.0% | unknown | 16 Jul 2026 | — | Source ↗ | |
| 6 | GLM 5.2 Open source Z.ai | 31.7% | max | 16 Jun 2026 | — | Source ↗ | |
| 7 | Claude Opus 4.7 (xhigh) Anthropic | 28.6% | xhigh | 16 Apr 2026 | — | Source ↗ | |
| 8 | GPT-5.5 (xhigh) OpenAI | 27.2% | xhigh | 23 Apr 2026 | — | Source ↗ | |
| 9 | Grok 4.5 SpaceXAI | 23.4% | high | 8 Jul 2026 | — | Source ↗ | |
| 10 | Gemini 3.1 Pro Preview Google | 22.0% | — | 19 Feb 2026 | — | Source ↗ | |
| 11 | GPT-5.4 OpenAI | 19.0% | high | 5 Mar 2026 | — | Source ↗ |
Caveats: settings (reasoning effort, agent scaffold) differ between rows and materially affect scores; the best reported setting per model is shown. Source: Epoch AI, CC BY 4.0.