| # | Model | Score | Relative | Setting | Released | Run | Evidence |
|---|---|---|---|---|---|---|---|
| 1 | Kimi K3 (unknown) Open source Moonshot | 87.0% | unknown | 16 Jul 2026 | — | Source ↗ | |
| 2 | Tencent Hy4 preview (unknown) Open source Tencent | 75.0% | unknown | 28 Aug 2026 | — | Source ↗ | |
| 3 | DeepSeek V4 Flash 0731 (unknown) Open source DeepSeek | 56.0% | unknown | 31 Jul 2026 | — | Source ↗ | |
| 4 | DeepSeek V4.1 Flash (unknown) Open source DeepSeek | 54.0% | unknown | 9 Sep 2026 | — | Source ↗ | |
| 5 | DeepSeek V4 Pro 0813 (unknown) Open source DeepSeek | 50.0% | unknown | 13 Aug 2026 | — | Source ↗ | |
| 6 | GLM 5.3 Open source Z.ai | 49.0% | max | 14 Aug 2026 | — | Source ↗ | |
| 7 | GLM 5.2 Open source Z.ai | 35.0% | max | 16 Jun 2026 | — | Source ↗ | |
| 8 | GLM 5.1 Open source Z.ai | 22.2% | — | 7 Apr 2026 | — | Source ↗ | |
| 9 | MiMo-V2.5-Pro Open source Xiaomi | 22.0% | — | 22 Apr 2026 | — | Source ↗ | |
| 10 | GLM 5.3 Flash Open source Z.ai | 21.0% | max | 20 Aug 2026 | — | Source ↗ | |
| 11 | MiniMax M3 Open source MiniMax | 18.0% | — | 1 Jun 2026 | — | Source ↗ | |
| 12 | Qwen3.8 27B Open source Qwen | 16.0% | — | 14 Aug 2026 | — | Source ↗ | |
| 13 | DeepSeek v4 Pro (high) Open source DeepSeek | 16.0% | max | 24 Apr 2026 | — | Source ↗ | |
| 14 | MiMo-V2.5 Open source Xiaomi | 16.0% | — | 22 Apr 2026 | — | Source ↗ | |
| 15 | Kimi K2.6 Open source MoonshotAI | 16.0% | — | 20 Apr 2026 | — | Source ↗ | |
| 16 | Mistral Medium 3.5 Open source Mistral | 9.0% | — | 28 Apr 2026 | — | Source ↗ | |
| 17 | DeepSeek-V3.2 (Thinking; Fireworks) Open source DeepSeek | 8.0% | — | 1 Dec 2025 | — | Source ↗ | |
| 18 | Inkling Small Open source Thinking Machines | 6.0% | — | 15 Jul 2026 | — | Source ↗ | |
| 19 | GLM 4.7 Open source Z.ai | 6.0% | — | 22 Dec 2025 | — | Source ↗ | |
| 20 | MiniMax M2.5 Open source MiniMax | 4.0% | — | 12 Feb 2026 | — | Source ↗ | |
| 21 | MiniMax M2.7 Open source MiniMax | 3.0% | — | 18 Mar 2026 | — | Source ↗ | |
| 22 | Nemotron 3 Ultra Open source NVIDIA | 2.0% | — | 4 Jun 2026 | — | Source ↗ | |
| 23 | Inkling Open source Thinking Machines | 0.0% | — | 15 Jul 2026 | — | Source ↗ | |
| 24 | Laguna XS.2 Open source Poolside | 0.0% | — | 28 Apr 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.