Updated daily at 05:00 UTC
Benchmark
FrontierMath Tiers 1-3 (v1)
Measures: The first private set of 290 expert-written problems from advanced undergraduate to early research mathematics, each with an automatically checkable answer; replaced by v2 on 2026-06-12.
- Maintainer
- Epoch AI
- Introduced
- 2024
- Status
- Superseded
- Saturation
- Saturated
- Typical provenance
- Independent
- In scoring set
- No
Leaderboard
Independent and third-party results only.
| Model | Accuracy | Provenance | Source | Evaluated |
|---|---|---|---|---|
| Gemini 1.5 Flash | 0.0% | Independent | Epoch AI | 7 Mar 2025 |
| Llama 4 Scout | 0.0% | Independent | Epoch AI | 8 Apr 2025 |
| Claude 3.5 Haiku | 0.3% | Independent | Epoch AI | 7 Mar 2025 |
| GPT-4o | 0.3% | Independent | Epoch AI | 7 Mar 2025 |
| GPT-4o | 0.3% | Independent | Epoch AI | 6 Mar 2025 |
| Mistral Large 2.1 | 0.3% | Independent | Epoch AI | 6 Mar 2025 |
| Mistral Medium 3 | 0.3% | Independent | Epoch AI | 7 May 2025 |
| Grok 2 | 0.7% | Independent | Epoch AI | 6 Mar 2025 |
| Llama 4 Maverick | 0.7% | Independent | Epoch AI | 8 Apr 2025 |
| Claude 3.5 Sonnet | 1.0% | Independent | Epoch AI | 7 Mar 2025 |
| GPT-4.1 nano | 1.0% | Independent | Epoch AI | 14 Apr 2025 |
| Qwen2.5-Max | 1.0% | Independent | Epoch AI | 2 Apr 2025 |
| o1-mini OpenAI · High effort | 1.4% | Independent | Epoch AI | 6 Mar 2025 |
| DeepSeek-V3 | 1.7% | Independent | Epoch AI | 7 Mar 2025 |
| Gemini 2.0 Flash | 1.7% | Independent | Epoch AI | 9 Mar 2025 |
| o1-mini OpenAI · Medium effort | 1.7% | Independent | Epoch AI | 6 Mar 2025 |
| Qwen-Plus | 1.7% | Independent | Epoch AI | 14 May 2025 |
| Claude 3.5 Sonnet | 2.1% | Independent | Epoch AI | 6 Mar 2025 |
| GPT-5.1 OpenAI · None effort | 2.1% | Independent | Epoch AI | 25 Nov 2025 |
| GLM-4.7 | 2.4% | Independent | Epoch AI | 30 Jan 2026 |
| Grok 3 Mini SpaceXAI (xAI) · Low effort | 2.8% | Independent | Epoch AI | 10 Apr 2025 |
| Claude 3.7 Sonnet Anthropic · 64K effort | 3.1% | Independent | Epoch AI | 13 Mar 2025 |
| Claude 3.7 Sonnet | 3.1% | Independent | Epoch AI | 6 Mar 2025 |
| Claude 3.7 Sonnet Anthropic · 32K effort | 3.4% | Independent | Epoch AI | 13 Mar 2025 |
| Grok 3 | 3.8% | Independent | Epoch AI | 10 Apr 2025 |
| GLM-4.6 | 3.8% | Independent | Epoch AI | 8 Dec 2025 |
| Claude 3.7 Sonnet Anthropic · 16K effort | 4.1% | Independent | Epoch AI | 6 Mar 2025 |
| Claude Haiku 4.5 | 4.1% | Independent | Epoch AI | 16 Oct 2025 |
| Claude Opus 4 Anthropic · 27K effort | 4.1% | Independent | Epoch AI | 5 Aug 2025 |
| Claude Sonnet 4 | 4.1% | Independent | Epoch AI | 4 Jul 2025 |
| Claude Opus 4 | 4.5% | Independent | Epoch AI | 4 Jul 2025 |
| GPT-4.1 mini | 4.5% | Independent | Epoch AI | 14 Apr 2025 |
| Gemini 2.5 Flash | 4.8% | Independent | Epoch AI | 18 Dec 2025 |
| GPT-4.1 | 5.5% | Independent | Epoch AI | 14 Apr 2025 |
| Claude Opus 4.1 | 5.9% | Independent | Epoch AI | 5 Aug 2025 |
| Grok 3 Mini SpaceXAI (xAI) · High effort | 5.9% | Independent | Epoch AI | 10 Apr 2025 |
| Claude Haiku 4.5 Anthropic · 32K effort | 5.9% | Independent | Epoch AI | 22 Oct 2025 |
| Qwen3.5-Flash | 6.2% | Independent | Epoch AI | 12 May 2026 |
| Claude Opus 4.1 Anthropic · 27K effort | 7.2% | Independent | Epoch AI | 5 Aug 2025 |
| GPT-5 nano OpenAI · Medium effort | 7.2% | Independent | Epoch AI | 7 Aug 2025 |
| o3-mini OpenAI · Medium effort | 8.1% | Independent | Epoch AI | 6 Mar 2025 |
| GPT-5 nano OpenAI · High effort | 8.3% | Independent | Epoch AI | 30 Oct 2025 |
| Qwen3-235B-A22B-Thinking-2507 | 8.5% | Independent | Epoch AI | 11 Dec 2025 |
| Claude Sonnet 4.5 | 9.3% | Independent | Epoch AI | 16 Nov 2025 |
| o1 OpenAI · High effort | 9.3% | Independent | Epoch AI | 7 Mar 2025 |
| o3 OpenAI · Low effort | 9.7% | Independent | Epoch AI | 17 Nov 2025 |
| Gemini 2.5 Pro | 10.3% | Independent | Epoch AI | 3 Jul 2025 |
| Qwen3.6-Flash | 10.3% | Independent | Epoch AI | 12 May 2026 |
| o4-mini OpenAI · Low effort | 10.7% | Independent | Epoch AI | 16 Nov 2025 |
| o3-mini OpenAI · High effort | 12.4% | Independent | Epoch AI | 16 Nov 2025 |
| Claude Sonnet 4.5 Anthropic · 59K effort | 13.5% | Independent | Epoch AI | 13 Nov 2025 |
| Gemini 2.5 Pro | 14.1% | Independent | Epoch AI | 24 Nov 2025 |
| Claude Sonnet 4.5 Anthropic · 32K effort | 15.2% | Independent | Epoch AI | 16 Nov 2025 |
| GLM-5 | 16.4% | Independent | Epoch AI | 19 Feb 2026 |
| o3 OpenAI · Medium effort | 16.9% | Independent | Epoch AI | 16 Nov 2025 |
| GPT-5.1 OpenAI · Low effort | 17.3% | Independent | Epoch AI | 25 Nov 2025 |
| o3 OpenAI · High effort | 18.7% | Independent | Epoch AI | 16 Nov 2025 |
| o4-mini OpenAI · Medium effort | 19.0% | Independent | Epoch AI | 13 Nov 2025 |
| Grok 4 | 19.7% | Independent | Epoch AI | 13 Nov 2025 |
| Claude Opus 4.5 Anthropic · 16K effort | 20.3% | Independent | Epoch AI | 25 Nov 2025 |
| GPT-5 mini OpenAI · Medium effort | 20.3% | Independent | Epoch AI | 13 Nov 2025 |
| Claude Opus 4.5 | 20.7% | Independent | Epoch AI | 25 Nov 2025 |
| Claude Opus 4.5 Anthropic · 32K effort | 20.7% | Independent | Epoch AI | 25 Nov 2025 |
| Qwen3.5-Plus | 21.0% | Independent | Epoch AI | 14 May 2026 |
| Kimi K2 Thinking | 21.4% | Independent | Epoch AI | 5 Dec 2025 |
| Qwen3.6-Max-Preview | 23.1% | Independent | Epoch AI | 27 May 2026 |
| o4-mini OpenAI · High effort | 24.8% | Independent | Epoch AI | 13 Nov 2025 |
| GPT-5.4 nano OpenAI · High effort | 25.9% | Independent | Epoch AI | 15 Apr 2026 |
| Qwen3.6-Plus | 26.2% | Independent | Epoch AI | 12 May 2026 |
| GPT-5.2 OpenAI · Low effort | 26.6% | Independent | Epoch AI | 11 Dec 2025 |
| GPT-5.1 OpenAI · Medium effort | 26.9% | Independent | Epoch AI | 17 Nov 2025 |
| GPT-5 OpenAI · Medium effort | 27.2% | Independent | Epoch AI | 13 Nov 2025 |
| GPT-5 mini OpenAI · High effort | 27.2% | Independent | Epoch AI | 13 Nov 2025 |
| GPT-5.4 mini OpenAI · High effort | 28.3% | Independent | Epoch AI | 15 Apr 2026 |
| GPT-5.1 OpenAI · High effort | 31.0% | Independent | Epoch AI | 13 Nov 2025 |
| Claude Sonnet 4.6 Anthropic · 16K effort | 32.4% | Independent | Epoch AI | 20 Feb 2026 |
| GPT-5 OpenAI · High effort | 32.4% | Independent | Epoch AI | 13 Nov 2025 |
| GLM-5.1 | 33.4% | Independent | Epoch AI | 11 May 2026 |
| Gemini 3 Flash | 35.6% | Independent | Epoch AI | 17 Dec 2025 |
| GPT-5.2 OpenAI · Medium effort | 36.9% | Independent | Epoch AI | 11 Dec 2025 |
| Gemini 3.1 Pro | 36.9% | Independent | Epoch AI | 19 Feb 2026 |
| Gemini 3 Pro | 37.6% | Independent | Epoch AI | 21 Nov 2025 |
| Claude Opus 4.6 | 38.3% | Independent | Epoch AI | 5 Feb 2026 |
| Gemini 3.5 Flash Google DeepMind · High effort | 39.0% | Independent | Epoch AI | 22 May 2026 |
| Kimi K2.6 | 39.0% | Independent | Epoch AI | 7 May 2026 |
| Claude Opus 4.6 Anthropic · 64K effort | 39.7% | Independent | Epoch AI | 6 Feb 2026 |
| Claude Opus 4.6 Anthropic · 32K effort | 40.0% | Independent | Epoch AI | 6 Feb 2026 |
| GPT-5.2 OpenAI · High effort | 40.3% | Independent | Epoch AI | 11 Dec 2025 |
| Claude Opus 4.6 Anthropic · Max effort | 40.7% | Independent | Epoch AI | 12 Feb 2026 |
| GPT-5.2 OpenAI · Xhigh effort | 40.7% | Independent | Epoch AI | 13 Dec 2025 |
| Claude Opus 4.7 Anthropic · Xhigh effort | 43.8% | Independent | Epoch AI | 17 Apr 2026 |
| Claude Opus 4.8 Anthropic · Max effort | 47.2% | Independent | Epoch AI | 8 Jun 2026 |
| GPT-5.4 OpenAI · Xhigh effort | 47.6% | Independent | Epoch AI | 6 Mar 2026 |
| GPT-5.4 Pro OpenAI · Xhigh effort | 50.0% | Independent | Epoch AI | 6 Mar 2026 |
- Epoch AI, 'Capabilities & benchmarking', epoch.ai/benchmarks, CC BY 4.0.