The hardest tier of FrontierMath: problems that would occupy a specialist researcher for days. Designed to stay unsaturated for years.
The absolute frontier of automated mathematical reasoning.
Small problem count, so a single solved problem moves the percentage a lot. Treat small gaps between models as noise.
| # | Model | Lab | Score | Measured | Method | Source |
|---|---|---|---|---|---|---|
| 1 | Claude Fable 5 | Anthropic | 100.0% | 2026-06-10 | independent | Epoch AI Benchmarking Hub |
| 2 | GPT-5.6 Sol | OpenAI | 81.7% | 2026-07-09 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 3 | AI Co-Mathematician | Google DeepMind | 75.6% | 2026-06-12 | independent | Epoch AI Benchmarking Hub |
| 4 | Claude Opus 5 | Anthropic | 73.2% | 2026-07-24 | independent | Epoch AI Benchmarking Hub |
| 5 | GPT-5.5 | OpenAI | 72.5% | 2026-06-11 | independent | Epoch AI Benchmarking Hub |
| 6 | GPT-5.6 Terra | OpenAI | 70.7% | 2026-07-09 | independent | Epoch AI Benchmarking Hub |
| 7 | GPT-5.6 Luna | OpenAI | 61.0% | 2026-07-09 | independent | Epoch AI Benchmarking Hub |
| 8 | GPT-5.4 Pro | OpenAI | 58.5% | 2026-06-13 | independent | Epoch AI Benchmarking Hub |
| 9 | Claude Opus 4.8 | Anthropic | 56.1% | 2026-06-10 | independent | Epoch AI Benchmarking Hub |
| 10 | GPT-5.4 | OpenAI | 50.0% | 2026-04-02 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 11 | Qwen 3.8 Max | Alibaba | 46.3% | 2026-08-04 | independent | Epoch AI Benchmarking Hub |
| 12 | GPT-5.2 Pro | OpenAI | 46.0% | 2026-06-13 | independent | Epoch AI Benchmarking Hub |
| 13 | GPT-5.5 Pro | OpenAI | 39.6% | 2026-04-23 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 14 | Kimi K3open | Moonshot AI | 39.0% | 2026-07-17 | independent | Epoch AI Benchmarking Hub |
| 15 | Gemini 3.7 Flash | Google DeepMind | 36.6% | 2026-08-14 | independent | Epoch AI Benchmarking Hub |
| 16 | Qwen3.7-Max | Alibaba | 34.1% | 2026-06-13 | independent | Epoch AI Benchmarking Hub |
| 17 | Claude Opus 4.7 | Anthropic | 31.7% | 2026-06-10 | independent | Epoch AI Benchmarking Hub |
| 18 | Grok 4.6 | xAI | 31.7% | 2026-08-14 | independent | Epoch AI Benchmarking Hub |
| 19 | Claude Sonnet 5 | Anthropic | 29.3% | 2026-06-30 | independent | Epoch AI Benchmarking Hub |
| 20 | GLM-5.2open | Z.ai | 29.3% | 2026-06-19 | independent | Epoch AI Benchmarking Hub |
| 21 | Gemini 3.1 Pro | Google DeepMind | 26.8% | 2026-06-11 | independent | Epoch AI Benchmarking Hub |
| 22 | Gemini 3.5 Flash | Google DeepMind | 26.8% | 2026-06-10 | independent | Epoch AI Benchmarking Hub |
| 23 | Kimi K2.6open | Moonshot AI | 25.6% | 2026-06-10 | independent | Epoch AI Benchmarking Hub |
| 24 | DeepSeek V4 Flash 0731open | DeepSeek | 24.4% | 2026-08-02 | independent | Epoch AI Benchmarking Hub |
| 25 | Grok 4.5 | xAI | 24.4% | 2026-07-09 | independent | Epoch AI Benchmarking Hub |
| 26 | Gemini 3.6 Flash | Google DeepMind | 22.0% | 2026-08-02 | independent | Epoch AI Benchmarking Hub |
| 27 | Claude Opus 4.6 | Anthropic | 19.8% | 2026-02-12 | independent · 4 runs | Epoch AI Benchmarking Hub |
| 28 | GPT-5 Pro | OpenAI | 19.5% | 2026-06-12 | independent | Epoch AI Benchmarking Hub |
| 29 | Gemini 3 Pro | Google DeepMind | 18.8% | 2025-11-21 | independent | Epoch AI Benchmarking Hub |
| 30 | Gemini 3 Flash | Google DeepMind | 17.1% | 2026-06-11 | independent | Epoch AI Benchmarking Hub |
| 31 | Grok 4.20 | xAI | 17.1% | 2026-07-13 | independent | Epoch AI Benchmarking Hub |
| 32 | Inkling-Smallopen | Thinking Machines | 17.1% | 2026-08-14 | independent | Epoch AI Benchmarking Hub |
| 33 | GPT-5.2 | OpenAI | 15.1% | 2025-12-14 | independent · 4 runs | Epoch AI Benchmarking Hub |
| 34 | Grok 4.3 Beta | xAI | 14.6% | 2026-06-17 | independent | Epoch AI Benchmarking Hub |
| 35 | Muse Spark | Meta AI | 14.6% | 2026-04-08 | independent | Epoch AI Benchmarking Hub |
| 36 | GPT-5.4 Nano | OpenAI | 12.2% | 2026-06-12 | independent | Epoch AI Benchmarking Hub |
| 37 | Kimi K2.7 Codeopen | Moonshot AI | 12.2% | 2026-06-13 | independent | Epoch AI Benchmarking Hub |
| 38 | Gemini 2.5 Deep Think | Google DeepMind | 10.4% | — | independent | Epoch AI Benchmarking Hub |
| 39 | GPT-5.4 Mini | OpenAI | 9.8% | 2026-06-12 | independent | Epoch AI Benchmarking Hub |
| 40 | GPT-5.1 | OpenAI | 8.3% | 2025-11-17 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 41 | Inklingopen | Thinking Machines | 4.9% | 2026-08-06 | independent | Epoch AI Benchmarking Hub |
| 42 | Gemini 2.5 Flash | Google DeepMind | 4.2% | 2025-12-18 | independent | Epoch AI Benchmarking Hub |
| 43 | Kimi K2.5open | Moonshot AI | 4.2% | 2026-02-02 | independent | Epoch AI Benchmarking Hub |
| 44 | Qwen 3.6 Max (Preview) | Alibaba | 4.2% | 2026-05-28 | independent | Epoch AI Benchmarking Hub |
| 45 | Claude Opus 4.5 | Anthropic | 3.5% | 2025-11-25 | independent · 3 runs | Epoch AI Benchmarking Hub |
| 46 | Claude Sonnet 4.5 | Anthropic | 3.1% | 2025-10-22 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 47 | Claude Opus 4.1 | Anthropic | 2.4% | 2026-06-11 | independent | Epoch AI Benchmarking Hub |
| 48 | DeepSeek-V4-Proopen | DeepSeek | 2.4% | 2026-06-17 | independent | Epoch AI Benchmarking Hub |
| 49 | GPT-5.5 Instant | OpenAI | 2.4% | 2026-08-02 | independent | Epoch AI Benchmarking Hub |
| 50 | Claude Haiku 4.5 | Anthropic | 2.1% | 2025-10-22 | independent | Epoch AI Benchmarking Hub |
| 51 | Claude Opus 4 | Anthropic | 2.1% | 2025-07-01 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 52 | DeepSeek-V3.2open | DeepSeek | 2.1% | 2025-12-16 | independent | Epoch AI Benchmarking Hub |
| 53 | GLM-4.6open | Z.ai | 2.1% | 2025-12-08 | independent | Epoch AI Benchmarking Hub |
| 54 | GLM-5open | Z.ai | 2.1% | 2026-02-19 | independent | Epoch AI Benchmarking Hub |
| 55 | Grok 4 Heavy | xAI | 2.1% | 2025-10-09 | independent | Epoch AI Benchmarking Hub |
| 56 | o3 | OpenAI | 2.1% | 2025-07-01 | independent | Epoch AI Benchmarking Hub |
| 57 | Qwen 3.5 Plus (hosted 397B-A17B) | Alibaba | 2.1% | 2026-05-15 | independent | Epoch AI Benchmarking Hub |
| 58 | Claude 3.5 Sonnet | Anthropic | 0.0% | 2025-07-01 | independent | Epoch AI Benchmarking Hub |
| 59 | Claude 3.5 Sonnet (October 2024) | Anthropic | 0.0% | 2025-07-01 | independent | Epoch AI Benchmarking Hub |
| 60 | Claude 3.7 Sonnet | Anthropic | 0.0% | 2025-07-01 | independent | Epoch AI Benchmarking Hub |
| 61 | Claude Sonnet 4 | Anthropic | 0.0% | 2025-07-01 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 62 | Claude Sonnet 4.6 | Anthropic | 0.0% | 2026-02-22 | independent | Epoch AI Benchmarking Hub |
| 63 | DeepSeek-R1 (May 2025)open | DeepSeek | 0.0% | 2025-07-01 | independent | Epoch AI Benchmarking Hub |
| 64 | Gemini 2.5 Pro (Jun 2025) | Google DeepMind | 0.0% | 2025-07-03 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 65 | Gemini 3.5 Flash-Lite | Google DeepMind | 0.0% | 2026-08-02 | independent | Epoch AI Benchmarking Hub |
| 66 | GLM-4.5open | Z.ai | 0.0% | 2025-09-08 | independent | Epoch AI Benchmarking Hub |
| 67 | GLM-4.7open | Z.ai | 0.0% | 2026-01-30 | independent | Epoch AI Benchmarking Hub |
| 68 | GLM-5.1open | Z.ai | 0.0% | 2026-05-12 | independent | Epoch AI Benchmarking Hub |
| 69 | GPT-4.1 | OpenAI | 0.0% | 2025-07-01 | independent | Epoch AI Benchmarking Hub |
| 70 | GPT-5 | OpenAI | 0.0% | 2025-10-30 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 71 | GPT-5 mini | OpenAI | 0.0% | 2025-10-30 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 72 | GPT-5 nano | OpenAI | 0.0% | 2025-10-30 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 73 | Grok 3 | xAI | 0.0% | 2025-07-01 | independent | Epoch AI Benchmarking Hub |
| 74 | Grok 4 | xAI | 0.0% | 2025-08-11 | independent | Epoch AI Benchmarking Hub |
| 75 | Grok-3 mini | xAI | 0.0% | 2025-07-01 | independent | Epoch AI Benchmarking Hub |
| 76 | Kimi K2 Thinkingopen | Moonshot AI | 0.0% | 2025-12-08 | independent | Epoch AI Benchmarking Hub |
| 77 | o3-mini | OpenAI | 0.0% | 2026-06-11 | independent | Epoch AI Benchmarking Hub |
| 78 | o4-mini | OpenAI | 0.0% | 2025-08-07 | independent · 2 runs | Epoch AI Benchmarking Hub |
| 79 | Qwen 3.5 Flash (hosted 35B-A3B) | Alibaba | 0.0% | 2026-05-12 | independent | Epoch AI Benchmarking Hub |
| 80 | Qwen 3.6 Flash | Alibaba | 0.0% | 2026-05-12 | independent | Epoch AI Benchmarking Hub |
| 81 | Qwen 3.6 Plus | Alibaba | 0.0% | 2026-05-12 | independent | Epoch AI Benchmarking Hub |
| 82 | Qwen3-235B-A22B (Jul 2025)open | Alibaba | 0.0% | 2025-12-11 | independent | Epoch AI Benchmarking Hub |