Updated daily at 05:00 UTC
Benchmark
FrontierMath Tiers 1-3
Measures: 295 unpublished, expert-written math problems (advanced undergraduate to graduate level) with automatically verifiable answers; a model may use Python under a 1M-token limit and gets 1 point per correct answer on the private set.
- Maintainer
- Epoch AI (OpenAI-funded; Epoch discloses the conflict of interest)
- Introduced
- 2024
- Status
- Active
- Saturation
- Saturated
- Typical provenance
- Independent
- In scoring set
- No
Leaderboard
Every result we hold, whoever ran it. Only independent and third-party rows count towards a score.
| Model | Accuracy | Provenance | Source | Evaluated |
|---|---|---|---|---|
| GPT-3.5 Turbo | 0.0% | Independent | Epoch AI | 27 Aug 2026 |
| GPT-4o | 0.4% | Independent | Epoch AI | 27 Aug 2026 |
| GPT-4 Turbo | 0.7% | Independent | Epoch AI | 27 Aug 2026 |
| GPT-4o mini | 0.7% | Independent | Epoch AI | 27 Aug 2026 |
| GPT-5 nano OpenAI · Minimal effort | 1.8% | Independent | Epoch AI | 27 Aug 2026 |
| o3-mini OpenAI · Low effort | 3.9% | Independent | Epoch AI | 27 Aug 2026 |
| GPT-5.4 nano OpenAI · None effort | 4.6% | Independent | Epoch AI | 28 Aug 2026 |
| GPT-4.1 | 6.0% | Independent | Epoch AI | 27 Aug 2026 |
| GPT-5 mini OpenAI · Minimal effort | 6.0% | Independent | Epoch AI | 27 Aug 2026 |
| GPT-5 nano OpenAI · Low effort | 6.0% | Independent | Epoch AI | 27 Aug 2026 |
| GPT-4.1 mini | 6.7% | Independent | Epoch AI | 27 Aug 2026 |
| o1 OpenAI · Low effort | 8.4% | Independent | Epoch AI | 27 Aug 2026 |
| Qwen3.5-Flash | 9.5% | Independent | Epoch AI | 28 Aug 2026 |
| o1 OpenAI · Medium effort | 10.2% | Independent | Epoch AI | 27 Aug 2026 |
| o3-mini OpenAI · Medium effort | 10.5% | Independent | Epoch AI | 27 Aug 2026 |
| Claude Opus 4.1 Anthropic · 32K effort | 12.6% | Independent | Epoch AI | 11 Jun 2026 |
| o1 OpenAI · High effort | 14.7% | Independent | Epoch AI | 28 Aug 2026 |
| o4-mini OpenAI · Low effort | 16.1% | Independent | Epoch AI | 27 Aug 2026 |
| GPT-5.4 mini OpenAI · None effort | 17.2% | Independent | Epoch AI | 28 Aug 2026 |
| Qwen3.6-Flash | 17.2% | Independent | Epoch AI | 29 Aug 2026 |
| Qwen3.6-35B-A3B | 17.5% | Independent | Epoch AI | 29 Aug 2026 |
| GPT-5 OpenAI · Minimal effort | 18.2% | Independent | Epoch AI | 27 Aug 2026 |
| GPT-5 mini OpenAI · Low effort | 18.2% | Independent | Epoch AI | 27 Aug 2026 |
| Qwen3.5-Flash Alibaba Group · None effort | 18.2% | Independent | Epoch AI | 28 Aug 2026 |
| o3-mini OpenAI · High effort | 18.6% | Independent | Epoch AI | 11 Jun 2026 |
| Qwen3-Max | 18.9% | Independent | Epoch AI | 30 Aug 2026 |
| o3 OpenAI · Low effort | 19.3% | Independent | Epoch AI | 27 Aug 2026 |
| Qwen3.7-Flash Alibaba Group · None effort | 19.3% | Independent | Epoch AI | 29 Aug 2026 |
| Qwen3.7-Flash | 19.3% | Independent | Epoch AI | 29 Aug 2026 |
| GPT-5 nano OpenAI · High effort | 20.0% | Independent | Epoch AI | 12 Jun 2026 |
| GPT-5.4 nano OpenAI · Low effort | 20.4% | Independent | Epoch AI | 28 Aug 2026 |
| Qwen3.6-35B-A3B Alibaba Group · None effort | 20.4% | Independent | Epoch AI | 30 Aug 2026 |
| Gemini 3.1 Flash-Lite Google DeepMind · Minimal effort | 21.4% | Independent | Epoch AI | 28 Aug 2026 |
| Gemini 3.1 Flash-Lite Google DeepMind · Low effort | 22.5% | Independent | Epoch AI | 30 Aug 2026 |
| Qwen3.6-Flash Alibaba Group · None effort | 22.5% | Independent | Epoch AI | 29 Aug 2026 |
| Claude Sonnet 4.5 Anthropic · 32K effort | 23.9% | Independent | Epoch AI | 11 Jun 2026 |
| Gemini 2.5 Pro | 24.6% | Independent | Epoch AI | 11 Jun 2026 |
| GPT-5.4 mini OpenAI · Low effort | 24.6% | Independent | Epoch AI | 28 Aug 2026 |
| GLM-5.1 Z.ai (Zhipu AI) · None effort | 24.9% | Independent | Epoch AI | 28 Aug 2026 |
| Gemini 3.5 Flash-Lite Google DeepMind · High effort | 26.0% | Independent | Epoch AI | 2 Aug 2026 |
| Gemini 3.1 Flash-Lite Google DeepMind · High effort | 27.7% | Independent | Epoch AI | 30 Aug 2026 |
| o4-mini OpenAI · Medium effort | 28.8% | Independent | Epoch AI | 27 Aug 2026 |
| Qwen3.5-397B-A17B | 29.5% | Independent | Epoch AI | 28 Aug 2026 |
| o3 OpenAI · Medium effort | 29.8% | Independent | Epoch AI | 27 Aug 2026 |
| Qwen3.5-397B-A17B Alibaba Group · None effort | 31.2% | Independent | Epoch AI | 30 Aug 2026 |
| Qwen3.6-Plus Alibaba Group · None effort | 32.3% | Independent | Epoch AI | 29 Aug 2026 |
| Inkling Thinking Machines Lab · Xhigh effort | 33.3% | Independent | Epoch AI | 6 Aug 2026 |
| o3 OpenAI · High effort | 33.3% | Independent | Epoch AI | 27 Aug 2026 |
| Qwen3.6-27B Alibaba Group · None effort | 34.0% | Independent | Epoch AI | 29 Aug 2026 |
| Claude Opus 4.5 Anthropic · 32K effort | 34.4% | Independent | Epoch AI | 11 Jun 2026 |
| Qwen3.7-Plus Alibaba Group · None effort | 34.4% | Independent | Epoch AI | 29 Aug 2026 |
| Qwen3.6-27B | 35.1% | Independent | Epoch AI | 28 Aug 2026 |
| o4-mini OpenAI · High effort | 36.1% | Independent | Epoch AI | 11 Jun 2026 |
| GLM-5.1 | 36.8% | Independent | Epoch AI | 29 Aug 2026 |
| GPT-5 OpenAI · Low effort | 37.2% | Independent | Epoch AI | 27 Aug 2026 |
| Qwen3.6-Plus | 38.2% | Independent | Epoch AI | 30 Aug 2026 |
| GPT-5.6 Luna OpenAI · None effort | 39.6% | Independent | Epoch AI | 29 Aug 2026 |
| GPT-5.6 Luna OpenAI · Low effort | 41.4% | Independent | Epoch AI | 29 Aug 2026 |
| GLM-5.2 Z.ai (Zhipu AI) · None effort | 42.5% | Independent | Epoch AI | 29 Aug 2026 |
| Grok 4.3 SpaceXAI (xAI) · High effort | 42.8% | Independent | Epoch AI | 17 Jun 2026 |
| GPT-5.4 nano OpenAI · High effort | 44.9% | Independent | Epoch AI | 12 Jun 2026 |
| Grok 4.20 | 44.9% | Independent | Epoch AI | 13 Jul 2026 |
| DeepSeek-V4-Pro DeepSeek · Max effort | 45.3% | Independent | Epoch AI | 17 Jun 2026 |
| Inkling Small Thinking Machines Lab · Xhigh effort | 46.3% | Independent | Epoch AI | 14 Aug 2026 |
| GPT-5 mini OpenAI · High effort | 46.7% | Independent | Epoch AI | 12 Jun 2026 |
| Gemini 3 Flash | 51.2% | Independent | Epoch AI | 11 Jun 2026 |
| GPT-5.4 mini OpenAI · Xhigh effort | 51.2% | Independent | Epoch AI | 12 Jun 2026 |
| Kimi K2.7 Code | 54.0% | Independent | Epoch AI | 13 Jun 2026 |
| GLM-5.2 Z.ai (Zhipu AI) · Low effort | 54.7% | Independent | Epoch AI | 29 Aug 2026 |
| GPT-5 OpenAI · High effort | 55.4% | Independent | Epoch AI | 10 Jun 2026 |
| GLM-5.3-Flash Z.ai (Zhipu AI) · Max effort | 55.8% | Independent | Epoch AI | 27 Aug 2026 |
| GPT-5 Pro OpenAI · High effort | 55.8% | Independent | Epoch AI | 12 Jun 2026 |
| Grok 4.5 SpaceXAI (xAI) · High effort | 57.2% | Independent | Epoch AI | 9 Jul 2026 |
| Kimi K2.6 | 57.2% | Independent | Epoch AI | 10 Jun 2026 |
| DeepSeek-V4-Flash DeepSeek · Max effort | 57.5% | Independent | Epoch AI | 2 Aug 2026 |
| Gemini 3.6 Flash Google DeepMind · High effort | 58.9% | Independent | Epoch AI | 2 Aug 2026 |
| GLM-5.2 Z.ai (Zhipu AI) · Max effort | 59.2% | Independent | Epoch AI | 19 Jun 2026 |
| Gemini 3.1 Pro | 59.6% | Independent | Epoch AI | 11 Jun 2026 |
| Gemini 3.5 Flash Google DeepMind · High effort | 62.8% | Independent | Epoch AI | 10 Jun 2026 |
| DeepSeek-V4-Pro DeepSeek · Max effort | 64.6% | Independent | Epoch AI | 19 Aug 2026 |
| Qwen3.7-Max | 64.6% | Independent | Epoch AI | 13 Jun 2026 |
| Claude Sonnet 5 Anthropic · Max effort | 65.6% | Independent | Epoch AI | 30 Jun 2026 |
| Qwen3.8-Max Alibaba Group · Xhigh effort | 65.6% | Independent | Epoch AI | 2 Sep 2026 |
| Claude Opus 4.6 Anthropic · Max effort | 66.0% | Independent | Epoch AI | 11 Jun 2026 |
| Grok 4.6 SpaceXAI (xAI) · Xhigh effort | 66.0% | Independent | Epoch AI | 14 Aug 2026 |
| GPT-5.2 OpenAI · Xhigh effort | 67.4% | Independent | Epoch AI | 11 Jun 2026 |
| Gemini 3.8 Flash Google DeepMind · High effort | 68.4% | Independent | Epoch AI | 2 Sep 2026 |
| GLM-5.3 Z.ai (Zhipu AI) · Max effort | 68.8% | Independent | Epoch AI | 25 Aug 2026 |
| Claude Opus 4.7 Anthropic · Max effort | 70.2% | Independent | Epoch AI | 10 Jun 2026 |
| Gemini 3.7 Flash Google DeepMind · High effort | 71.6% | Independent | Epoch AI | 14 Aug 2026 |
| Kimi K3 Moonshot AI · Max effort | 72.2% | Independent | Epoch AI | 17 Jul 2026 |
| GPT-5.2 Pro OpenAI · Xhigh effort | 74.0% | Independent | Epoch AI | 13 Jun 2026 |
| Muse Spark 1.3 Meta · Max effort | 74.0% | Independent | Epoch AI | 18 Sep 2026 |
| Muse Spark 1.3 Meta · Xhigh effort | 74.4% | Independent | Epoch AI | 16 Sep 2026 |
| Qwen3.8-Max Alibaba Group · Xhigh effort | 74.7% | Independent | Epoch AI | 4 Aug 2026 |
| GPT-5.4 OpenAI · Xhigh effort | 78.6% | Independent | Epoch AI | 11 Jun 2026 |
| Claude Opus 4.8 Anthropic · Max effort | 80.0% | Independent | Epoch AI | 10 Jun 2026 |
| GPT-5.6 Luna OpenAI · Max effort | 82.1% | Independent | Epoch AI | 9 Jul 2026 |
| GPT-5.4 Pro OpenAI · Xhigh effort | 82.5% | Independent | Epoch AI | 13 Jun 2026 |
| GPT-5.5 OpenAI · Xhigh effort | 85.3% | Independent | Epoch AI | 11 Jun 2026 |
| Claude Opus 5 Anthropic · Max effort | 85.6% | Independent | Epoch AI | 24 Jul 2026 |
| GPT-5.6 Terra OpenAI · Max effort | 86.0% | Independent | Epoch AI | 9 Jul 2026 |
| Claude Fable 5 Anthropic · Max effort | 87.0% | Independent | Epoch AI | 9 Jun 2026 |
| GPT-5.5 Pro OpenAI · Xhigh effort | 87.7% | Independent | Epoch AI | 12 Jun 2026 |
| GPT-5.6 Sol OpenAI · Max effort | 89.1% | Independent | Epoch AI | 9 Jul 2026 |
| Claude Fable 5.1 Anthropic · Max effort | 90.2% | Independent | Epoch AI | 1 Sep 2026 |
| Claude Opus 5.5 Anthropic · Max effort | 91.2% | Independent | Epoch AI | 22 Sep 2026 |
| GPT-6 Astra OpenAI · Max effort | 93.7% | Independent | Epoch AI | 30 Aug 2026 |
- Epoch AI, 'Capabilities & benchmarking', epoch.ai/benchmarks, CC BY 4.0.