math benchmark · included in ECI
FrontierMath-Tiers-1-3-v2-Private
Unpublished, expert-written research-level mathematics problems (tiers 1-3).
Results
15
Random baseline
0.0%
Score ceiling
100%
Released
12 Jun 2026
Score by model release date
Best score by organisation
Leaderboard
| # | Model | Score | Relative | Setting | Released | Run | Evidence |
|---|---|---|---|---|---|---|---|
| 1 | Kimi K3 Open source MoonshotAI | 72.2% ±2.7 | max | 16 Jul 2026 | 17 Jul 2026 | Source ↗ | |
| 2 | GLM 5.3 Open source Z.ai | 68.8% ±2.7 | max | 14 Aug 2026 | 25 Aug 2026 | Source ↗ | |
| 3 | DeepSeek V4 Pro 0813 Open source DeepSeek | 64.6% ±2.8 | max | 13 Aug 2026 | 19 Aug 2026 | Log ↗ | |
| 4 | GLM 5.2 Open source Z.ai | 59.2% ±3.0 | max | 16 Jun 2026 | 19 Jun 2026 | Source ↗ | |
| 5 | DeepSeek V4 Flash 0731 Open source DeepSeek | 57.5% ±2.9 | max | 31 Jul 2026 | 2 Aug 2026 | Log ↗ | |
| 6 | Kimi K2.6 Open source MoonshotAI | 57.2% ±2.9 | — | 20 Apr 2026 | 10 Jun 2026 | Log ↗ | |
| 7 | GLM 5.3 Flash Open source Z.ai | 55.8% ±2.9 | max | 20 Aug 2026 | 27 Aug 2026 | Source ↗ | |
| 8 | Kimi K2.7 Code Open source MoonshotAI | 54.0% ±3.0 | — | 12 Jun 2026 | 13 Jun 2026 | Log ↗ | |
| 9 | Inkling Small (xhigh) Open source Thinking Machines | 46.3% ±3.0 | xhigh | 15 Jul 2026 | 14 Aug 2026 | Log ↗ | |
| 10 | DeepSeek v4 Pro (high) Open source DeepSeek | 45.3% ±3.0 | max | 24 Apr 2026 | 17 Jun 2026 | Log ↗ | |
| 11 | GLM 5.1 Open source Z.ai | 36.8% ±2.9 | — | 7 Apr 2026 | 29 Aug 2026 | Log ↗ | |
| 12 | Qwen3.6 27B Open source Qwen | 35.1% ±2.8 | — | 22 Apr 2026 | 28 Aug 2026 | Log ↗ | |
| 13 | Inkling (xhigh) Open source Thinking Machines | 33.3% ±2.8 | xhigh | 15 Jul 2026 | 6 Aug 2026 | Log ↗ | |
| 14 | Qwen3.5 397B A17B Open source Qwen | 31.2% ±2.7 | none | 13 Feb 2026 | 30 Aug 2026 | Log ↗ | |
| 15 | Qwen3.6 35B A3B Open source Qwen | 20.4% ±2.4 | none | 27 Apr 2026 | 30 Aug 2026 | Log ↗ |
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.