FrontierMath-Tiers-1-3-v2-Private
Re-run by Epoch AITiers 1 to 3 of FrontierMath (second version): unpublished mathematics problems written for Epoch AI, from advanced undergraduate to research level.
- Kind
- Run independently by Epoch AI, with public logs. Our own re-runs are not built yet.
- Score shown
- Epoch's column "Best score (across scorers)", as a percentage, higher is better; random guessing scores about 0.0%. Error bars as the source gives them.
- Results
- 127 in all: 26 for open-weight models, 97 for closed ones, 4 where Epoch does not say. 25 linked to a model page here.
Best open-weight result: moonshotai/Kimi-K3.1-Pro, 78.2%, number 25 of 127 results.
| # | Model | Weights | Setting | Score | Date | Source |
|---|---|---|---|---|---|---|
| 1 | claude-fable-5-2_maxAnthropic | closed | effort max | 95.1%±1.3 | 6 Nov 2026 | Epoch AI |
| 2 | gpt-6-astra_maxOpenAI | closed | effort max | 93.7%±1.4 | 30 Aug 2026 | Epoch AI |
| 3 | gpt-6.1-sol_maxOpenAI | closed | effort max | 93.7%±1.4 | 29 Sep 2026 | Epoch AI |
| 4 | gpt-6.2-astra_maxOpenAI | closed | effort max | 92.6%±1.6 | 27 Oct 2026 | Epoch AI |
| 5 | claude-opus-5-5_maxAnthropic | closed | effort max | 91.2%±1.7 | 22 Sep 2026 | Epoch AI |
| 6 | gpt-6.1-terra_maxOpenAI | closed | effort max | 90.9%±1.7 | 26 Oct 2026 | Epoch AI |
| 7 | gemini-4-carbon_highGoogle DeepMind | closed | effort high | 90.5%±1.7 | 2 Nov 2026 | Epoch AI |
| 8 | claude-fable-5-1_maxAnthropic | closed | effort max | 90.2%±1.8 | 1 Sep 2026 | Epoch AI |
| 9 | gpt-6-sol_maxOpenAI | closed | effort max | 89.8%±1.8 | 22 Sep 2026 | Epoch AI |
| 10 | gpt-5.6-sol_maxOpenAI | closed | effort max | 89.1%±1.8 | 9 Jul 2026 | Epoch AI log |
| 11 | claude-sonnet-5-5_maxAnthropic | closed | effort max | 88.8%±1.9 | 29 Sep 2026 | Epoch AI |
| 12 | gemini-4-hydrogen_xhighGoogle DeepMind | closed | effort xhigh | 88.1%±1.9 | 13 Nov 2026 | Epoch AI |
| 13 | gpt-5.5-pro_xhighOpenAI | closed | effort xhigh | 87.7%±1.9 | 12 Jun 2026 | Epoch AI |
| 14 | claude-fable-5_maxAnthropic | closed | effort max | 87.0%±2.0 | 9 Jun 2026 | Epoch AI log |
| 15 | gemini-4-argon_highGoogle DeepMind | closed | effort high | 86.0%±2.1 | 13 Oct 2026 | Epoch AI |
| 16 | gpt-5.6-terra_maxOpenAI | closed | effort max | 86.0%±2.1 | 9 Jul 2026 | Epoch AI log |
| 17 | claude-opus-5_maxAnthropic | closed | effort max | 85.6%±2.1 | 24 Jul 2026 | Epoch AI log |
| 18 | gpt-5.5_xhighOpenAI | closed | effort xhigh | 85.3%±2.1 | 11 Jun 2026 | Epoch AI |
| 19 | gpt-5.4-pro-2026-03-05_xhighOpenAI | closed | effort xhigh | 82.5%±2.3 | 13 Jun 2026 | Epoch AI log |
| 20 | gpt-5.6-luna_maxOpenAI | closed | effort max | 82.1%±2.3 | 9 Jul 2026 | Epoch AI log |
| 21 | muse-wonder-1.0_maxMeta AI | closed | effort max | 82.1%±2.3 | 12 Oct 2026 | Epoch AI |
| 22 | claude-opus-4-8_maxAnthropic | closed | effort max | 80.0%±2.4 | 10 Jun 2026 | Epoch AI log |
| 23 | gpt-6-luna_maxOpenAI | closed | effort max | 78.9%±2.4 | 22 Sep 2026 | Epoch AI |
| 24 | gpt-5.4-2026-03-05_xhighOpenAI | closed | effort xhigh | 78.6%±2.4 | 11 Jun 2026 | Epoch AI log |
| 25 | moonshotai/Kimi-K3.1-ProEpoch: kimi-k3.1-pro-together_max, Moonshot | open | effort max | 78.2%±2.4 | 6 Nov 2026 | Epoch AI |
| 26 | claude-haiku-5-5_maxAnthropic | closed | effort max | 75.1%±2.6 | 7 Oct 2026 | Epoch AI |
| 27 | qwen3.8-max_xhighAlibaba | closed | effort xhigh | 74.7%±2.6 | 4 Aug 2026 | Epoch AI log |
| 28 | muse-spark-1.3_xhighMeta AI | closed | effort xhigh | 74.4%±2.6 | 16 Sep 2026 | Epoch AI |
| 29 | muse-spark-1.3_maxMeta AI | closed | effort max | 74.0%±2.6 | 18 Sep 2026 | Epoch AI |
| 30 | gpt-5.2-pro-2025-12-11_xhighOpenAI | closed | effort xhigh | 74.0%±2.6 | 13 Jun 2026 | Epoch AI |
| 31 | moonshotai/Kimi-K3Epoch: kimi-k3_max, Moonshot | open | effort max | 72.2%±2.7 | 17 Jul 2026 | Epoch AI |
| 32 | gemini-3.7-flash_highGoogle DeepMind | closed | effort high | 71.6%±2.7 | 14 Aug 2026 | Epoch AI log |
| 33 | deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_max, DeepSeek | open | effort max | 71.2%±2.7 | 19 Oct 2026 | Epoch AI |
| 34 | claude-opus-4-7_maxAnthropic | closed | effort max | 70.2%±2.7 | 10 Jun 2026 | Epoch AI log |
| 35 | meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_xhigh, Meta AI | open | effort xhigh | 69.8%±2.7 | 15 Oct 2026 | Epoch AI |
| 36 | zai-org/GLM-5.3Epoch: glm-5.3_max, Z.ai (Zhipu AI) | open | effort max | 68.8%±2.7 | 25 Aug 2026 | Epoch AI |
| 37 | gemini-3.8-flash_highGoogle DeepMind | closed | effort high | 68.4%±2.8 | 2 Sep 2026 | Epoch AI |
| 38 | gpt-5.2-2025-12-11_xhighOpenAI | closed | effort xhigh | 67.4%±2.8 | 11 Jun 2026 | Epoch AI |
| 39 | deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_max, DeepSeek | open | effort max | 67.4%±2.8 | 8 Oct 2026 | Epoch AI |
| 40 | claude-opus-4-6_maxAnthropic | closed | effort max | 66.0%±2.8 | 11 Jun 2026 | Epoch AI log |
| 41 | grok-4.6_xhighxAI | closed | effort xhigh | 66.0%±2.8 | 14 Aug 2026 | Epoch AI log |
| 42 | claude-sonnet-5_maxAnthropic | closed | effort max | 65.6%±2.8 | 30 Jun 2026 | Epoch AI log |
| 43 | qwen3.8-max-0902_xhighAlibaba | closed | effort xhigh | 65.6%±2.8 | 2 Sep 2026 | Epoch AI |
| 44 | deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_max, DeepSeek | open | effort max | 64.6%±2.8 | 19 Aug 2026 | Epoch AI log |
| 45 | qwen3.7-maxAlibaba | closed | default | 64.6%±2.8 | 13 Jun 2026 | Epoch AI log |
| 46 | gemini-3.5-flash_highGoogle DeepMind | closed | effort high | 62.8%±2.9 | 10 Jun 2026 | Epoch AI log |
| 47 | gemini-3.1-pro-previewGoogle DeepMind | closed | default | 59.6%±2.9 | 11 Jun 2026 | Epoch AI log |
| 48 | zai-org/GLM-5.2Epoch: glm-5.2_max, Z.ai (Zhipu AI) | open | effort max | 59.2%±3.0 | 19 Jun 2026 | Epoch AI |
| 49 | gemini-3.6-flash_highGoogle DeepMind | closed | effort high | 58.9%±2.9 | 2 Aug 2026 | Epoch AI log |
| 50 | mistral-large-4-0_highMistral AIno page match | open | effort high | 58.6%±2.9 | 12 Oct 2026 | Epoch AI |
| 51 | deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_max, DeepSeek | open | effort max | 57.5%±2.9 | 2 Aug 2026 | Epoch AI log |
| 52 | grok-4.5_highxAI | closed | effort high | 57.2%±2.9 | 9 Jul 2026 | Epoch AI |
| 53 | moonshotai/Kimi-K2.6Epoch: kimi-k2.6, Moonshot | open | default | 57.2%±2.9 | 10 Jun 2026 | Epoch AI log |
| 54 | zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_max, Z.ai (Zhipu AI) | open | effort max | 55.8%±2.9 | 27 Aug 2026 | Epoch AI |
| 55 | gpt-5-pro-2025-10-06_highOpenAI | closed | effort high | 55.8%±2.9 | 12 Jun 2026 | Epoch AI log |
| 56 | gpt-5-2025-08-07_highOpenAI | closed | effort high | 55.4%±2.9 | 10 Jun 2026 | Epoch AI |
| 57 | zai-org/GLM-5.2Epoch: glm-5.2_low, Z.ai (Zhipu AI) | open | effort low | 54.7%±3.0 | 29 Aug 2026 | Epoch AI log |
| 58 | moonshotai/Kimi-K2.7-CodeEpoch: kimi-k2.7-code, Moonshot | open | default | 54.0%±3.0 | 13 Jun 2026 | Epoch AI log |
| 59 | grok-4.7_xhighxAI | closed | effort xhigh | 53.0%±3.0 | 22 Sep 2026 | Epoch AI |
| 60 | gemini-3-flash-previewGoogle DeepMind | closed | default | 51.2%±3.0 | 11 Jun 2026 | Epoch AI log |
| 61 | gpt-5.4-mini-2026-03-17_xhighOpenAI | closed | effort xhigh | 51.2%±3.0 | 12 Jun 2026 | Epoch AI |
| 62 | gpt-5-mini-2025-08-07_highOpenAI | closed | effort high | 46.7%±3.0 | 12 Jun 2026 | Epoch AI log |
| 63 | thinkingmachines/Inkling-SmallEpoch: inkling-small_xhigh, Thinking Machines | open | effort xhigh | 46.3%±3.0 | 14 Aug 2026 | Epoch AI log |
| 64 | deepseek-ai/DeepSeek-V4-ProEpoch: deepseek-v4-pro_max, DeepSeek | open | effort max | 45.3%±3.0 | 17 Jun 2026 | Epoch AI log |
| 65 | gpt-5.4-nano-2026-03-17_highOpenAI | closed | effort high | 44.9%±3.0 | 12 Jun 2026 | Epoch AI log |
| 66 | grok-4.20-0309-reasoningxAI | closed | default | 44.9%±3.0 | 13 Jul 2026 | Epoch AI log |
| 67 | grok-4.3_highxAI | closed | effort high | 42.8%±2.9 | 17 Jun 2026 | Epoch AI log |
| 68 | zai-org/GLM-5.2Epoch: glm-5.2_none, Z.ai (Zhipu AI) | open | effort none | 42.5%±2.9 | 29 Aug 2026 | Epoch AI log |
| 69 | gpt-5.6-luna_lowOpenAI | closed | effort low | 41.4%±2.9 | 29 Aug 2026 | Epoch AI log |
| 70 | gpt-5.6-luna_noneOpenAI | closed | effort none | 39.6%±2.9 | 29 Aug 2026 | Epoch AI log |
| 71 | qwen3.6-plusAlibaba | closed | default | 38.2%±2.9 | 30 Aug 2026 | Epoch AI log |
| 72 | gpt-5-2025-08-07_lowOpenAI | closed | effort low | 37.2%±2.9 | 27 Aug 2026 | Epoch AI log |
| 73 | zai-org/GLM-5.1Epoch: glm-5.1, Z.ai (Zhipu AI) | open | default | 36.8%±2.9 | 29 Aug 2026 | Epoch AI log |
| 74 | o4-mini-2025-04-16_highOpenAI | closed | effort high | 36.1%±2.9 | 11 Jun 2026 | Epoch AI log |
| 75 | Qwen/Qwen3.6-27BEpoch: qwen3.6-27b, Alibaba | open | default | 35.1%±2.8 | 28 Aug 2026 | Epoch AI log |
| 76 | claude-opus-4-5-20251101_32KAnthropic | closed | default | 34.4%±2.8 | 11 Jun 2026 | Epoch AI log |
| 77 | qwen3.7-plus_noneAlibaba | closed | effort none | 34.4%±2.8 | 29 Aug 2026 | Epoch AI log |
| 78 | Qwen/Qwen3.6-27BEpoch: qwen3.6-27b_none, Alibaba | open | effort none | 34.0%±2.8 | 29 Aug 2026 | Epoch AI log |
| 79 | thinkingmachines/InklingEpoch: Inkling_xhigh, Thinking Machines | open | effort xhigh | 33.3%±2.8 | 6 Aug 2026 | Epoch AI log |
| 80 | o3-2025-04-16_highOpenAI | closed | effort high | 33.3%±2.8 | 27 Aug 2026 | Epoch AI log |
| 81 | qwen3.6-plus_noneAlibaba | closed | effort none | 32.3%±2.8 | 29 Aug 2026 | Epoch AI log |
| 82 | Qwen/Qwen3.5-397B-A17BEpoch: qwen3.5-397b-a17b_none, Alibaba | open | effort none | 31.2%±2.7 | 30 Aug 2026 | Epoch AI log |
| 83 | o3-2025-04-16_mediumOpenAI | closed | effort medium | 29.8%±2.7 | 27 Aug 2026 | Epoch AI log |
| 84 | Qwen/Qwen3.5-397B-A17BEpoch: qwen3.5-397b-a17b, Alibaba | open | default | 29.5%±2.7 | 28 Aug 2026 | Epoch AI log |
| 85 | o4-mini-2025-04-16_mediumOpenAI | closed | effort medium | 28.8%±2.7 | 27 Aug 2026 | Epoch AI log |
| 86 | gemini-3.1-flash-lite_highGoogle | closed | effort high | 27.7%±2.7 | 30 Aug 2026 | Epoch AI log |
| 87 | gpt-5.5-instantOpenAI | closed | default | 26.3%±2.6 | 2 Aug 2026 | Epoch AI log |
| 88 | gemini-3.5-flash-lite_highGoogle DeepMind | closed | effort high | 26.0%±2.6 | 2 Aug 2026 | Epoch AI log |
| 89 | zai-org/GLM-5.1Epoch: glm-5.1_none, Z.ai (Zhipu AI) | open | effort none | 24.9%±2.6 | 28 Aug 2026 | Epoch AI log |
| 90 | gemini-2.5-proGoogle DeepMind | closed | default | 24.6%±2.6 | 11 Jun 2026 | Epoch AI log |
| 91 | gpt-5.4-mini-2026-03-17_lowOpenAI | closed | effort low | 24.6%±2.6 | 28 Aug 2026 | Epoch AI log |
| 92 | claude-sonnet-4-5-20250929_32KAnthropic | closed | default | 23.9%±2.5 | 11 Jun 2026 | Epoch AI log |
| 93 | gemini-3.1-flash-lite_lowGoogle | closed | effort low | 22.5%±2.5 | 30 Aug 2026 | Epoch AI log |
| 94 | qwen3.6-flash_noneAlibaba | not stated | effort none | 22.5%±2.5 | 29 Aug 2026 | Epoch AI log |
| 95 | gemini-3.1-flash-lite_minimalGoogle | closed | effort minimal | 21.4%±2.4 | 28 Aug 2026 | Epoch AI log |
| 96 | gpt-5.4-nano-2026-03-17_lowOpenAI | closed | effort low | 20.4%±2.4 | 28 Aug 2026 | Epoch AI log |
| 97 | Qwen/Qwen3.6-35B-A3BEpoch: qwen3.6-35b-a3b_none, Alibaba | open | effort none | 20.4%±2.4 | 30 Aug 2026 | Epoch AI log |
| 98 | gpt-5-nano-2025-08-07_highOpenAI | closed | effort high | 20.0%±2.4 | 12 Jun 2026 | Epoch AI log |
| 99 | o3-2025-04-16_lowOpenAI | closed | effort low | 19.3%±2.3 | 27 Aug 2026 | Epoch AI log |
| 100 | qwen3.7-flashAlibaba | closed | default | 19.3%±2.3 | 29 Aug 2026 | Epoch AI log |
| 101 | qwen3.7-flash_noneAlibaba | closed | effort none | 19.3%±2.3 | 29 Aug 2026 | Epoch AI log |
| 102 | qwen3-max-2025-09-23Alibaba | closed | default | 18.9%±2.3 | 30 Aug 2026 | Epoch AI log |
| 103 | o3-mini-2025-01-31_highOpenAI | closed | effort high | 18.6%±2.3 | 11 Jun 2026 | Epoch AI log |
| 104 | gpt-5-2025-08-07_minimalOpenAI | closed | effort minimal | 18.2%±2.3 | 27 Aug 2026 | Epoch AI log |
| 105 | gpt-5-mini-2025-08-07_lowOpenAI | closed | effort low | 18.2%±2.3 | 27 Aug 2026 | Epoch AI log |
| 106 | qwen3.5-flash_noneAlibaba | not stated | effort none | 18.2%±2.3 | 28 Aug 2026 | Epoch AI log |
| 107 | Qwen/Qwen3.6-35B-A3BEpoch: qwen3.6-35b-a3b, Alibaba | open | default | 17.5%±2.3 | 29 Aug 2026 | Epoch AI log |
| 108 | gpt-5.4-mini-2026-03-17_noneOpenAI | closed | effort none | 17.2%±2.2 | 28 Aug 2026 | Epoch AI log |
| 109 | qwen3.6-flashAlibaba | not stated | default | 17.2%±2.2 | 29 Aug 2026 | Epoch AI log |
| 110 | o4-mini-2025-04-16_lowOpenAI | closed | effort low | 16.1%±2.2 | 27 Aug 2026 | Epoch AI log |
| 111 | o1-2024-12-17_highOpenAI | closed | effort high | 14.7%±2.1 | 28 Aug 2026 | Epoch AI log |
| 112 | claude-opus-4-1-20250805_32KAnthropic | closed | default | 12.6%±2.0 | 11 Jun 2026 | Epoch AI log |
| 113 | o3-mini-2025-01-31_mediumOpenAI | closed | effort medium | 10.5%±1.8 | 27 Aug 2026 | Epoch AI log |
| 114 | o1-2024-12-17_mediumOpenAI | closed | effort medium | 10.2%±1.8 | 27 Aug 2026 | Epoch AI log |
| 115 | qwen3.5-flashAlibaba | not stated | default | 9.5%±1.7 | 28 Aug 2026 | Epoch AI log |
| 116 | o1-2024-12-17_lowOpenAI | closed | effort low | 8.4%±1.6 | 27 Aug 2026 | Epoch AI log |
| 117 | gpt-4.1-mini-2025-04-14OpenAI | closed | default | 6.7%±1.5 | 27 Aug 2026 | Epoch AI log |
| 118 | gpt-4.1-2025-04-14OpenAI | closed | default | 6.0%±1.4 | 27 Aug 2026 | Epoch AI log |
| 119 | gpt-5-mini-2025-08-07_minimalOpenAI | closed | effort minimal | 6.0%±1.4 | 27 Aug 2026 | Epoch AI log |
| 120 | gpt-5-nano-2025-08-07_lowOpenAI | closed | effort low | 6.0%±1.4 | 27 Aug 2026 | Epoch AI log |
| 121 | gpt-5.4-nano-2026-03-17_noneOpenAI | closed | effort none | 4.6%±1.2 | 28 Aug 2026 | Epoch AI log |
| 122 | o3-mini-2025-01-31_lowOpenAI | closed | effort low | 3.9%±1.1 | 27 Aug 2026 | Epoch AI log |
| 123 | gpt-5-nano-2025-08-07_minimalOpenAI | closed | effort minimal | 1.8%±0.8 | 27 Aug 2026 | Epoch AI log |
| 124 | gpt-4-turbo-2024-04-09OpenAI | closed | default | 0.7%±0.5 | 27 Aug 2026 | Epoch AI log |
| 125 | gpt-4o-mini-2024-07-18OpenAI | closed | default | 0.7%±0.5 | 27 Aug 2026 | Epoch AI log |
| 126 | gpt-4o-2024-08-06OpenAI | closed | default | 0.4%±0.4 | 27 Aug 2026 | Epoch AI log |
| 127 | gpt-3.5-turbo-0125OpenAI | closed | default | 0.0% | 27 Aug 2026 | Epoch AI log |
Open-weight rows are shaded. Dates are when the result was run or evaluated, where the data says. Data from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Epoch AI. Method