| # | Model | Score | Relative | Setting | Released | Run | Evidence |
|---|---|---|---|---|---|---|---|
| 1 | GPT-6 Astra OpenAI | 76.2% ±8.4 | max | 3 Sep 2026 | 30 Aug 2026 | Source ↗ | |
| 2 | Claude Opus 5.5 Anthropic | 71.4% ±7.1 | max | 22 Sep 2026 | 22 Sep 2026 | Source ↗ | |
| 3 | Claude Fable 5.1 Anthropic | 57.1% ±6.0 | max | 1 Sep 2026 | 8 Sep 2026 | Source ↗ | |
| 4 | GPT-6 Sol OpenAI | 53.3% ±4.6 | max | 22 Sep 2026 | 22 Sep 2026 | Source ↗ | |
| 5 | Claude Opus 5 Anthropic | 45.7% ±3.6 | max | 24 Jul 2026 | 5 Sep 2026 | Source ↗ | |
| 6 | GPT-5.6 Sol OpenAI | 44.8% ±4.2 | max | 9 Jul 2026 | 4 Sep 2026 | Source ↗ | |
| 7 | Claude Fable 5 Anthropic | 39.5% ±2.0 | max | 9 Jun 2026 | 17 Jul 2026 | Source ↗ | |
| 8 | GPT-5.5 (xhigh) OpenAI | 34.3% ±2.0 | xhigh | 23 Apr 2026 | 27 Jul 2026 | Source ↗ | |
| 9 | Grok 4.6 (xhigh) xAI | 30.5% ±1.6 | xhigh | 12 Aug 2026 | 18 Aug 2026 | Source ↗ | |
| 10 | Claude Opus 4.8 Anthropic | 28.6% ±1.4 | max | 28 May 2026 | 7 Aug 2026 | Source ↗ | |
| 11 | GPT-5.4 (xhigh) OpenAI | 25.4% | xhigh | 5 Mar 2026 | 25 Jun 2026 | Source ↗ | |
| 12 | GPT-5.2 (xhigh) OpenAI | 23.0% | xhigh | 11 Dec 2025 | 26 Jun 2026 | Source ↗ | |
| 13 | Claude Opus 4.7 Anthropic | 19.0% | max | 16 Apr 2026 | 30 Jun 2026 | Source ↗ | |
| 14 | Gemini 3.1 Pro Preview Google | 14.3% | — | 19 Feb 2026 | 25 Jun 2026 | Source ↗ | |
| 15 | Claude Opus 4.5 (128k thinking) Anthropic | 14.3% | 128K | 24 Nov 2025 | 25 Jun 2026 | Source ↗ | |
| 16 | Claude Opus 4.6 Anthropic | 12.7% | max | 5 Feb 2026 | 29 Jun 2026 | Source ↗ | |
| 17 | GPT-5 OpenAI | 12.7% | high | 7 Aug 2025 | 29 Jun 2026 | Source ↗ | |
| 18 | GLM 5.2 Open source Z.ai | 9.5% | max | 16 Jun 2026 | 29 Jun 2026 | Source ↗ | |
| 19 | Qwen3.7 Max Qwen | 9.5% | — | 19 May 2026 | 25 Jun 2026 | Source ↗ | |
| 20 | Claude Opus 4.1 Anthropic | 7.9% | — | 5 Aug 2025 | 25 Jun 2026 | Source ↗ | |
| 21 | Gemini 3.5 Flash Google | 4.8% | high | 19 May 2026 | 25 Jun 2026 | Source ↗ | |
| 22 | Kimi K2.6 Open source MoonshotAI | 2.4% | — | 20 Apr 2026 | 25 Jun 2026 | Source ↗ | |
| 23 | Claude Sonnet 4.5 (no thinking) Anthropic | 2.4% | — | 29 Sep 2025 | 25 Jun 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.