ARC-AGI-2
ReportedThe second, harder set of ARC-AGI grid puzzles: from a few examples the model must find the rule and apply it to a new grid.
- Kind
- Published by a third party's leaderboard, collected by Epoch AI. The leaderboard is named and linked.
- Score shown
- Epoch's column "Score", as a percentage, higher is better; random guessing scores about 0.0%. The source gives no error bars.
- Results
- 299 in all: 53 for open-weight models, 244 for closed ones, 2 where Epoch does not say. 49 linked to a model page here.
Best open-weight result: thinkingmachines/Inkling-2-Small, 81.7%, number 53 of 299 results.
| # | Model | Weights | Setting | Score | Date | Source |
|---|---|---|---|---|---|---|
| 1 | gpt-6-astra_maxOpenAI | closed | effort max | 95.0% | not dated | via Epoch AI |
| 2 | gemini-4-carbon_mediumGoogle DeepMind | closed | effort medium | 94.2% | not dated | via Epoch AI |
| 3 | gpt-6.1-sol_maxOpenAI | closed | effort max | 94.2% | not dated | via Epoch AI |
| 4 | claude-fable-5-2_xhighAnthropic | closed | effort xhigh | 93.8% | not dated | via Epoch AI |
| 5 | claude-opus-5-5_highAnthropic | closed | effort high | 93.3% | not dated | via Epoch AI |
| 6 | gpt-6-astra_xhighOpenAI | closed | effort xhigh | 93.3% | not dated | via Epoch AI |
| 7 | claude-fable-5-2_maxAnthropic | closed | effort max | 92.5% | not dated | via Epoch AI |
| 8 | claude-opus-5-5_xhighAnthropic | closed | effort xhigh | 92.5% | not dated | via Epoch AI |
| 9 | gpt-5.6-sol_maxOpenAI | closed | effort max | 92.5% | not dated | via Epoch AI |
| 10 | gemini-4-carbon_highGoogle DeepMind | closed | effort high | 92.1% | not dated | via Epoch AI |
| 11 | gpt-6-astra_highOpenAI | closed | effort high | 92.1% | not dated | via Epoch AI |
| 12 | gpt-6-astra_mediumOpenAI | closed | effort medium | 92.1% | not dated | via Epoch AI |
| 13 | claude-opus-5-5_maxAnthropic | closed | effort max | 91.7% | not dated | via Epoch AI |
| 14 | gpt-6.1-sol_highOpenAI | closed | effort high | 91.7% | not dated | via Epoch AI |
| 15 | gpt-6.1-sol_xhighOpenAI | closed | effort xhigh | 91.7% | not dated | via Epoch AI |
| 16 | gemini-4-argon_highGoogle DeepMind | closed | effort high | 91.3% | not dated | via Epoch AI |
| 17 | claude-opus-5_maxAnthropic | closed | effort max | 90.4% | not dated | via Epoch AI |
| 18 | claude-fable-5-2_highAnthropic | closed | effort high | 90.4% | not dated | via Epoch AI |
| 19 | claude-fable-5-1_maxAnthropic | closed | effort max | 90.0% | not dated | via Epoch AI |
| 20 | claude-fable-5-1_xhighAnthropic | closed | effort xhigh | 90.0% | not dated | via Epoch AI |
| 21 | gpt-5.6-sol_xhighOpenAI | closed | effort xhigh | 90.0% | not dated | via Epoch AI |
| 22 | gpt-5.6-sol_xhighOpenAI | closed | effort xhigh | 90.0% | not dated | via Epoch AI |
| 23 | claude-fable-5-2_mediumAnthropic | closed | effort medium | 89.6% | not dated | via Epoch AI |
| 24 | gpt-6-sol_maxOpenAI | closed | effort max | 89.6% | not dated | via Epoch AI |
| 25 | claude-fable-5_maxAnthropic | closed | effort max | 89.2% | not dated | via Epoch AI |
| 26 | gemini-3.8-flash_highGoogle DeepMind | closed | effort high | 89.2% | not dated | via Epoch AI |
| 27 | claude-fable-5-1_highAnthropic | closed | effort high | 88.8% | not dated | via Epoch AI |
| 28 | claude-fable-5_xhighAnthropic | closed | effort xhigh | 88.3% | not dated | via Epoch AI |
| 29 | claude-opus-5_highAnthropic | closed | effort high | 88.3% | not dated | via Epoch AI |
| 30 | claude-opus-5_maxAnthropic | closed | effort max | 88.3% | not dated | via Epoch AI |
| 31 | claude-fable-5_highAnthropic | closed | effort high | 87.5% | not dated | via Epoch AI |
| 32 | claude-opus-5-5_mediumAnthropic | closed | effort medium | 87.5% | not dated | via Epoch AI |
| 33 | gemini-4-argon_mediumGoogle DeepMind | closed | effort medium | 87.1% | not dated | via Epoch AI |
| 34 | claude-fable-5-2_lowAnthropic | closed | effort low | 86.7% | not dated | via Epoch AI |
| 35 | gpt-6.1-sol_mediumOpenAI | closed | effort medium | 86.7% | not dated | via Epoch AI |
| 36 | claude-fable-5-1_mediumAnthropic | closed | effort medium | 86.3% | not dated | via Epoch AI |
| 37 | gpt-5.6-sol_highOpenAI | closed | effort high | 85.4% | not dated | via Epoch AI |
| 38 | gpt-6-astra_lowOpenAI | closed | effort low | 85.4% | not dated | via Epoch AI |
| 39 | gpt-5.5_xhighOpenAI | closed | effort xhigh | 85.0% | not dated | via Epoch AI |
| 40 | gpt-5.5_xhighOpenAI | closed | effort xhigh | 85.0% | not dated | via Epoch AI |
| 41 | gemini-3.7-flash_highGoogle DeepMind | closed | effort high | 84.6% | not dated | via Epoch AI |
| 42 | gemini-3-deep-think-previewGoogle DeepMind | closed | default | 84.6% | not dated | via Epoch AI |
| 43 | gpt-5.5-pro_highOpenAI | closed | effort high | 84.6% | not dated | via Epoch AI |
| 44 | gpt-5.5-pro_xhighOpenAI | closed | effort xhigh | 84.2% | not dated | via Epoch AI |
| 45 | gpt-5.5-pro_xhighOpenAI | closed | effort xhigh | 84.2% | not dated | via Epoch AI |
| 46 | gpt-5.6-terra_maxOpenAI | closed | effort max | 83.9% | not dated | via Epoch AI |
| 47 | claude-sonnet-5-5_xhighAnthropic | closed | effort xhigh | 83.9% | not dated | via Epoch AI |
| 48 | gpt-5.4-pro-2026-03-05_xhighOpenAI | closed | effort xhigh | 83.3% | not dated | via Epoch AI |
| 49 | gpt-5.4-pro-2026-03-05_xhighOpenAI | closed | effort xhigh | 83.3% | not dated | via Epoch AI |
| 50 | gpt-5.5_highOpenAI | closed | effort high | 83.3% | not dated | via Epoch AI |
| 51 | gemini-3.8-flash_mediumGoogle DeepMind | closed | effort medium | 82.9% | not dated | via Epoch AI |
| 52 | claude-fable-5_mediumAnthropic | closed | effort medium | 82.5% | not dated | via Epoch AI |
| 53 | thinkingmachines/Inkling-2-SmallEpoch: inkling-2-small_xhigh, Thinking Machines | open | effort xhigh | 81.7% | not dated | via Epoch AI |
| 54 | claude-fable-5-1_lowAnthropic | closed | effort low | 78.3% | not dated | via Epoch AI |
| 55 | deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_max, DeepSeek | open | effort max | 78.1% | not dated | via Epoch AI |
| 56 | gpt-6-sol_xhighOpenAI | closed | effort xhigh | 78.1% | not dated | via Epoch AI |
| 57 | gemini-3.8-flash_lowGoogle DeepMind | closed | effort low | 77.5% | not dated | via Epoch AI |
| 58 | gemini-3.1-pro-previewGoogle DeepMind | closed | default | 77.1% | not dated | via Epoch AI |
| 59 | gemini-3.1-pro-previewGoogle DeepMind | closed | default | 77.1% | not dated | via Epoch AI |
| 60 | claude-fable-5_lowAnthropic | closed | effort low | 76.8% | not dated | via Epoch AI |
| 61 | gpt-6.1-sol_lowOpenAI | closed | effort low | 76.7% | not dated | via Epoch AI |
| 62 | claude-opus-4-7_maxAnthropic | closed | effort max | 75.8% | not dated | via Epoch AI |
| 63 | claude-opus-4-7_maxAnthropic | closed | effort max | 75.8% | not dated | via Epoch AI |
| 64 | gpt-5.6-terra_xhighOpenAI | closed | effort xhigh | 74.2% | not dated | via Epoch AI |
| 65 | gpt-5.6-terra_xhighOpenAI | closed | effort xhigh | 74.2% | not dated | via Epoch AI |
| 66 | gpt-5.4-2026-03-05_xhighOpenAI | closed | effort xhigh | 74.0% | not dated | via Epoch AI |
| 67 | gpt-5.4-2026-03-05_xhighOpenAI | closed | effort xhigh | 74.0% | not dated | via Epoch AI |
| 68 | deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_low, DeepSeek | open | effort low | 73.6% | not dated | via Epoch AI |
| 69 | deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_max, DeepSeek | open | effort max | 72.9% | not dated | via Epoch AI |
| 70 | claude-opus-4-8_highAnthropic | closed | effort high | 72.1% | not dated | via Epoch AI |
| 71 | gemini-3.5-flash_highGoogle DeepMind | closed | effort high | 72.1% | not dated | via Epoch AI |
| 72 | claude-opus-4-8_mediumAnthropic | closed | effort medium | 71.7% | not dated | via Epoch AI |
| 73 | gpt-5.5_mediumOpenAI | closed | effort medium | 70.4% | not dated | via Epoch AI |
| 74 | claude-opus-5-5_lowAnthropic | closed | effort low | 70.1% | not dated | via Epoch AI |
| 75 | claude-opus-4-6_120KAnthropic | closed | default | 69.2% | not dated | via Epoch AI |
| 76 | gpt-6-sol_highOpenAI | closed | effort high | 68.9% | not dated | via Epoch AI |
| 77 | claude-opus-4-6_120KAnthropic | closed | default | 68.8% | not dated | via Epoch AI |
| 78 | deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_high, DeepSeek | open | effort high | 68.6% | not dated | via Epoch AI |
| 79 | claude-opus-4-7_highAnthropic | closed | effort high | 68.3% | not dated | via Epoch AI |
| 80 | claude-opus-4-7_highAnthropic | closed | effort high | 68.3% | not dated | via Epoch AI |
| 81 | claude-opus-4-7_mediumAnthropic | closed | effort medium | 67.5% | not dated | via Epoch AI |
| 82 | deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_high, DeepSeek | open | effort high | 67.5% | not dated | via Epoch AI |
| 83 | gpt-5.4-2026-03-05_highOpenAI | closed | effort high | 67.5% | not dated | via Epoch AI |
| 84 | claude-sonnet-5-5_highAnthropic | closed | effort high | 67.4% | not dated | via Epoch AI |
| 85 | grok-4.6_xhighxAI | closed | effort xhigh | 67.1% | not dated | via Epoch AI |
| 86 | gpt-5.6-sol_mediumOpenAI | closed | effort medium | 67.1% | not dated | via Epoch AI |
| 87 | gpt-5.6-terra_highOpenAI | closed | effort high | 67.1% | not dated | via Epoch AI |
| 88 | claude-opus-4-6_120KAnthropic | closed | default | 66.3% | not dated | via Epoch AI |
| 89 | zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_max, Z.ai (Zhipu AI) | open | effort max | 65.8% | not dated | via Epoch AI |
| 90 | grok-4-20xAI | closed | default | 65.1% | not dated | via Epoch AI |
| 91 | grok-4.6_highxAI | closed | effort high | 65.1% | not dated | via Epoch AI |
| 92 | claude-opus-4-6_120KAnthropic | closed | default | 64.6% | not dated | via Epoch AI |
| 93 | gemini-3.7-flash_mediumGoogle DeepMind | closed | effort medium | 63.7% | not dated | via Epoch AI |
| 94 | claude-opus-4-8_lowAnthropic | closed | effort low | 62.2% | not dated | via Epoch AI |
| 95 | claude-opus-4-7_lowAnthropic | closed | effort low | 62.1% | not dated | via Epoch AI |
| 96 | deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_max, DeepSeek | open | effort max | 61.4% | not dated | via Epoch AI |
| 97 | grok-4.7_xhighxAI | closed | effort xhigh | 61.4% | not dated | via Epoch AI |
| 98 | claude-haiku-5-5_xhighAnthropic | closed | effort xhigh | 61.3% | not dated | via Epoch AI |
| 99 | deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_max, DeepSeek | open | effort max | 61.3% | not dated | via Epoch AI |
| 100 | grok-4.6_mediumxAI | closed | effort medium | 61.3% | not dated | via Epoch AI |
| 101 | deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_low, DeepSeek | open | effort low | 60.6% | not dated | via Epoch AI |
| 102 | claude-sonnet-4-6_highAnthropic | closed | effort high | 60.4% | not dated | via Epoch AI |
| 103 | gemini-3.6-flash_highGoogle DeepMind | closed | effort high | 60.4% | not dated | via Epoch AI |
| 104 | moonshotai/Kimi-K3Epoch: kimi-k3_max, Moonshot | open | effort max | 60.4% | not dated | via Epoch AI |
| 105 | deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_high, DeepSeek | open | effort high | 59.7% | not dated | via Epoch AI |
| 106 | gpt-6-astra_noneOpenAI | closed | effort none | 59.6% | not dated | via Epoch AI |
| 107 | gpt-5.6-luna_maxOpenAI | closed | effort max | 59.5% | not dated | via Epoch AI |
| 108 | gpt-6-luna_maxOpenAI | closed | effort max | 59.3% | not dated | via Epoch AI |
| 109 | grok-4.7_mediumxAI | closed | effort medium | 58.8% | not dated | via Epoch AI |
| 110 | grok-4.7_highxAI | closed | effort high | 58.3% | not dated | via Epoch AI |
| 111 | claude-sonnet-4-6_maxAnthropic | closed | effort max | 58.3% | not dated | via Epoch AI |
| 112 | gpt-6-sol_mediumOpenAI | closed | effort medium | 57.8% | not dated | via Epoch AI |
| 113 | meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_xhigh, Meta AI | open | effort xhigh | 57.5% | not dated | via Epoch AI |
| 114 | deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_low, DeepSeek | open | effort low | 56.3% | not dated | via Epoch AI |
| 115 | deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_high, DeepSeek | open | effort high | 56.0% | not dated | via Epoch AI |
| 116 | gpt-5.4-2026-03-05_mediumOpenAI | closed | effort medium | 55.4% | not dated | via Epoch AI |
| 117 | moonshotai/Kimi-K3Epoch: kimi-k3_high, Moonshot | open | effort high | 55.0% | not dated | via Epoch AI |
| 118 | gpt-5.2-pro-2025-12-11_highOpenAI | closed | effort high | 54.2% | not dated | via Epoch AI |
| 119 | gemini-3.7-flash_lowGoogle DeepMind | closed | effort low | 52.9% | not dated | via Epoch AI |
| 120 | gpt-5.2-2025-12-11_xhighOpenAI | closed | effort xhigh | 52.9% | not dated | via Epoch AI |
| 121 | gpt-5.2-2025-12-11_xhighOpenAI | closed | effort xhigh | 52.9% | not dated | via Epoch AI |
| 122 | gpt-5.2-2025-12-11_xhighOpenAI | closed | effort xhigh | 52.9% | not dated | via Epoch AI |
| 123 | grok-4.5_highxAI | closed | effort high | 52.6% | not dated | via Epoch AI |
| 124 | grok-4.5_mediumxAI | closed | effort medium | 52.6% | not dated | via Epoch AI |
| 125 | gemini-3.6-flash_mediumGoogle DeepMind | closed | effort medium | 50.4% | not dated | via Epoch AI |
| 126 | zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_high, Z.ai (Zhipu AI) | open | effort high | 50.1% | not dated | via Epoch AI |
| 127 | gemini-4-carbon_lowGoogle DeepMind | closed | effort low | 49.6% | not dated | via Epoch AI |
| 128 | gpt-5.6-luna_xhighOpenAI | closed | effort xhigh | 47.6% | not dated | via Epoch AI |
| 129 | gpt-5.6-luna_xhighOpenAI | closed | effort xhigh | 47.6% | not dated | via Epoch AI |
| 130 | meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_high, Meta AI | open | effort high | 47.5% | not dated | via Epoch AI |
| 131 | deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_low, DeepSeek | open | effort low | 46.0% | not dated | via Epoch AI |
| 132 | gemini-3-deep-think-previewGoogle DeepMind | closed | default | 45.1% | not dated | via Epoch AI |
| 133 | gpt-5.2-2025-12-11_highOpenAI | closed | effort high | 43.3% | not dated | via Epoch AI |
| 134 | gpt-5.6-sol_lowOpenAI | closed | effort low | 42.5% | not dated | via Epoch AI |
| 135 | Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_xhigh, Alibaba | open | effort xhigh | 42.4% | not dated | via Epoch AI |
| 136 | gemini-4-argon_lowGoogle DeepMind | closed | effort low | 42.1% | not dated | via Epoch AI |
| 137 | gpt-6-luna_xhighOpenAI | closed | effort xhigh | 41.9% | not dated | via Epoch AI |
| 138 | thinkingmachines/Inkling-SmallEpoch: inkling-small_xhigh, Thinking Machines | open | effort xhigh | 40.1% | not dated | via Epoch AI |
| 139 | gpt-5.2-pro-2025-12-11_mediumOpenAI | closed | effort medium | 38.5% | not dated | via Epoch AI |
| 140 | claude-opus-4-5-20251101_64KAnthropic | closed | default | 37.6% | not dated | via Epoch AI |
| 141 | gpt-5.6-terra_mediumOpenAI | closed | effort medium | 37.5% | not dated | via Epoch AI |
| 142 | thinkingmachines/InklingEpoch: Inkling, Thinking Machines | open | default | 36.5% | not dated | via Epoch AI |
| 143 | gemini-3-flash-previewGoogle DeepMind | closed | default | 33.6% | not dated | via Epoch AI |
| 144 | gpt-5.5_lowOpenAI | closed | effort low | 33.3% | not dated | via Epoch AI |
| 145 | grok-4.5_lowxAI | closed | effort low | 33.1% | not dated | via Epoch AI |
| 146 | thinkingmachines/Inkling-SmallEpoch: inkling-small_high, Thinking Machines | open | effort high | 33.1% | not dated | via Epoch AI |
| 147 | claude-sonnet-5-5_mediumAnthropic | closed | effort medium | 32.8% | not dated | via Epoch AI |
| 148 | gpt-6-sol_lowOpenAI | closed | effort low | 31.5% | not dated | via Epoch AI |
| 149 | gpt-6-luna_highOpenAI | closed | effort high | 31.4% | not dated | via Epoch AI |
| 150 | gemini-3-pro-previewGoogle DeepMind | closed | default | 31.1% | not dated | via Epoch AI |
| 151 | claude-haiku-5-5_highAnthropic | closed | effort high | 30.7% | not dated | via Epoch AI |
| 152 | gemini-3.6-flash_lowGoogle DeepMind | closed | effort low | 30.4% | not dated | via Epoch AI |
| 153 | gpt-5.6-luna_highOpenAI | closed | effort high | 29.3% | not dated | via Epoch AI |
| 154 | gpt-5.4-2026-03-05_lowOpenAI | closed | effort low | 29.2% | not dated | via Epoch AI |
| 155 | zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_low, Z.ai (Zhipu AI) | open | effort low | 27.9% | not dated | via Epoch AI |
| 156 | grok-4.6_lowxAI | closed | effort low | 27.6% | not dated | via Epoch AI |
| 157 | meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_medium, Meta AI | open | effort medium | 27.5% | not dated | via Epoch AI |
| 158 | gpt-5.2-2025-12-11_mediumOpenAI | closed | effort medium | 26.7% | not dated | via Epoch AI |
| 159 | claude-sonnet-5-5_lowAnthropic | closed | effort low | 24.0% | not dated | via Epoch AI |
| 160 | beam-1.0_xhighReflectionno page match | open | effort xhigh | 23.5% | not dated | via Epoch AI |
| 161 | claude-opus-4-5-20251101_16KAnthropic | closed | default | 22.8% | not dated | via Epoch AI |
| 162 | Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_low, Alibaba | open | effort low | 22.8% | not dated | via Epoch AI |
| 163 | zai-org/GLM-5.2Epoch: glm-5.2_unknown, Z.ai (Zhipu AI) | open | effort unknown | 22.8% | not dated | via Epoch AI |
| 164 | beam-1.0_highReflectionno page match | open | effort high | 20.0% | not dated | via Epoch AI |
| 165 | gpt-5.4-mini-2026-03-17_xhighOpenAI | closed | effort xhigh | 18.9% | not dated | via Epoch AI |
| 166 | gpt-5.4-mini-2026-03-17_xhighOpenAI | closed | effort xhigh | 18.9% | not dated | via Epoch AI |
| 167 | gpt-5.6-terra_lowOpenAI | closed | effort low | 18.8% | not dated | via Epoch AI |
| 168 | gpt-5-pro-2025-10-06_unknownOpenAI | closed | effort unknown | 18.3% | not dated | via Epoch AI |
| 169 | gpt-6-luna_mediumOpenAI | closed | effort medium | 18.1% | not dated | via Epoch AI |
| 170 | gpt-5.1-2025-11-13_highOpenAI | closed | effort high | 17.6% | not dated | via Epoch AI |
| 171 | grok-4.7_lowxAI | closed | effort low | 16.7% | not dated | via Epoch AI |
| 172 | grok-4-0709xAI | closed | default | 16.0% | not dated | via Epoch AI |
| 173 | claude-opus-4-5-20251101_8KAnthropic | closed | default | 13.9% | not dated | via Epoch AI |
| 174 | thinkingmachines/Inkling-SmallEpoch: inkling-small_medium, Thinking Machines | open | effort medium | 13.6% | not dated | via Epoch AI |
| 175 | claude-sonnet-4-5-20250929_32KAnthropic | closed | default | 13.6% | not dated | via Epoch AI |
| 176 | Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_medium, Alibaba | open | effort medium | 13.2% | not dated | via Epoch AI |
| 177 | gpt-5.4-mini-2026-03-17_highOpenAI | closed | effort high | 13.2% | not dated | via Epoch AI |
| 178 | gemini-3-flash-previewGoogle DeepMind | closed | default | 12.8% | not dated | via Epoch AI |
| 179 | moonshotai/Kimi-K3Epoch: kimi-k3_low, Moonshot | open | effort low | 12.4% | not dated | via Epoch AI |
| 180 | meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_low, Meta AI | open | effort low | 12.2% | not dated | via Epoch AI |
| 181 | moonshotai/Kimi-K2.5Epoch: kimi-k2.5, Moonshot | open | default | 11.8% | not dated | via Epoch AI |
| 182 | gemini-3.5-flash-lite_highGoogle DeepMind | closed | effort high | 10.3% | not dated | via Epoch AI |
| 183 | gpt-5-2025-08-07_highOpenAI | closed | effort high | 9.9% | not dated | via Epoch AI |
| 184 | gpt-5.2-2025-12-11_lowOpenAI | closed | effort low | 9.7% | not dated | via Epoch AI |
| 185 | claude-haiku-5-5_mediumAnthropic | closed | effort medium | 9.4% | not dated | via Epoch AI |
| 186 | gemini-3.5-flash_minimalGoogle DeepMind | closed | effort minimal | 8.9% | not dated | via Epoch AI |
| 187 | claude-opus-4-20250514_16KAnthropic | closed | default | 8.6% | not dated | via Epoch AI |
| 188 | claude-opus-4-5-20251101Anthropic | closed | default | 7.8% | not dated | via Epoch AI |
| 189 | gpt-5-2025-08-07_mediumOpenAI | closed | effort medium | 7.5% | not dated | via Epoch AI |
| 190 | gpt-5.6-luna_mediumOpenAI | closed | effort medium | 7.4% | not dated | via Epoch AI |
| 191 | claude-sonnet-4-5-20250929_16KAnthropic | closed | default | 6.9% | not dated | via Epoch AI |
| 192 | claude-sonnet-4-5-20250929_8KAnthropic | closed | default | 6.9% | not dated | via Epoch AI |
| 193 | gpt-5.1-2025-11-13_mediumOpenAI | closed | effort medium | 6.5% | not dated | via Epoch AI |
| 194 | o3-2025-04-16_highOpenAI | closed | effort high | 6.5% | not dated | via Epoch AI |
| 195 | beam-1.0_mediumReflectionno page match | open | effort medium | 6.4% | not dated | via Epoch AI |
| 196 | tiny-recursion-model | not stated | default | 6.3% | not dated | via Epoch AI |
| 197 | o4-mini-2025-04-16_highOpenAI | closed | effort high | 6.1% | not dated | via Epoch AI |
| 198 | claude-sonnet-4-20250514_16KAnthropic | closed | default | 5.9% | not dated | via Epoch AI |
| 199 | claude-sonnet-4-5-20250929_1KAnthropic | closed | default | 5.8% | not dated | via Epoch AI |
| 200 | gpt-5.4-nano-2026-03-17_xhighOpenAI | closed | effort xhigh | 5.7% | not dated | via Epoch AI |
| 201 | gpt-5.4-nano-2026-03-17_xhighOpenAI | closed | effort xhigh | 5.7% | not dated | via Epoch AI |
| 202 | grok-4-fastxAI | closed | default | 5.3% | not dated | via Epoch AI |
| 203 | gemini-3.5-flash-lite_mediumGoogle DeepMind | closed | effort medium | 5.3% | not dated | via Epoch AI |
| 204 | gpt-5.6-luna_lowOpenAI | closed | effort low | 5.1% | not dated | via Epoch AI |
| 205 | thinkingmachines/Inkling-SmallEpoch: inkling-small_low, Thinking Machines | open | effort low | 4.9% | not dated | via Epoch AI |
| 206 | gemini-2.5-pro_32KGoogle DeepMind | closed | default | 4.9% | not dated | via Epoch AI |
| 207 | zai-org/GLM-5Epoch: glm-5, Z.ai (Zhipu AI) | open | default | 4.9% | not dated | via Epoch AI |
| 208 | MiniMaxAI/MiniMax-M2.5Epoch: MiniMax-M2.5, MiniMax | open | default | 4.9% | not dated | via Epoch AI |
| 209 | o3-pro-2025-06-10_highOpenAI | closed | effort high | 4.9% | not dated | via Epoch AI |
| 210 | gpt-6-luna_lowOpenAI | closed | effort low | 4.6% | not dated | via Epoch AI |
| 211 | claude-opus-4-20250514_8KAnthropic | closed | default | 4.5% | not dated | via Epoch AI |
| 212 | gpt-5-mini-2025-08-07_highOpenAI | closed | effort high | 4.4% | not dated | via Epoch AI |
| 213 | gpt-5.4-mini-2026-03-17_mediumOpenAI | closed | effort medium | 4.4% | not dated | via Epoch AI |
| 214 | claude-haiku-4-5-20251001_32KAnthropic | closed | default | 4.0% | not dated | via Epoch AI |
| 215 | deepseek-ai/DeepSeek-V3.2Epoch: deepseek/deepseek-v3.2, DeepSeek | open | default | 4.0% | not dated | via Epoch AI |
| 216 | gemini-2.5-pro_16KGoogle DeepMind | closed | default | 4.0% | not dated | via Epoch AI |
| 217 | gpt-5-mini-2025-08-07_mediumOpenAI | closed | effort medium | 4.0% | not dated | via Epoch AI |
| 218 | thinkingmachines/Inkling-SmallEpoch: inkling-small_minimal, Thinking Machines | open | effort minimal | 4.0% | not dated | via Epoch AI |
| 219 | claude-sonnet-4-5-20250929Anthropic | closed | default | 3.8% | not dated | via Epoch AI |
| 220 | gpt-5.4-nano-2026-03-17_highOpenAI | closed | effort high | 3.6% | not dated | via Epoch AI |
| 221 | gemini-3-flash-previewGoogle DeepMind | closed | default | 3.3% | not dated | via Epoch AI |
| 222 | o3-mini-2025-01-31_highOpenAI | closed | effort high | 3.0% | not dated | via Epoch AI |
| 223 | o3-2025-04-16_mediumOpenAI | closed | effort medium | 3.0% | not dated | via Epoch AI |
| 224 | gemini-2.5-pro_8KGoogle DeepMind | closed | default | 2.9% | not dated | via Epoch AI |
| 225 | claude-haiku-4-5-20251001_16KAnthropic | closed | default | 2.8% | not dated | via Epoch AI |
| 226 | beam-1.0_lowReflectionno page match | open | effort low | 2.8% | not dated | via Epoch AI |
| 227 | claude-haiku-5-5_lowAnthropic | closed | effort low | 2.8% | not dated | via Epoch AI |
| 228 | gemini-3.6-flash_minimalGoogle DeepMind | closed | effort minimal | 2.6% | not dated | via Epoch AI |
| 229 | gpt-5-nano-2025-08-07_highOpenAI | closed | effort high | 2.6% | not dated | via Epoch AI |
| 230 | gemini-2.5-flash-preview-05-20_23KGoogle DeepMind | closed | default | 2.5% | not dated | via Epoch AI |
| 231 | o4-mini-2025-04-16_mediumOpenAI | closed | effort medium | 2.4% | not dated | via Epoch AI |
| 232 | gemini-2.5-flash-preview-05-20_1KGoogle DeepMind | closed | default | 2.2% | not dated | via Epoch AI |
| 233 | claude-sonnet-4-20250514_8KAnthropic | closed | default | 2.1% | not dated | via Epoch AI |
| 234 | gemini-2.5-flash-preview-05-20_8KGoogle DeepMind | closed | default | 2.1% | not dated | via Epoch AI |
| 235 | deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_none, DeepSeek | open | effort none | 2.1% | not dated | via Epoch AI |
| 236 | deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_none, DeepSeek | open | effort none | 2.1% | not dated | via Epoch AI |
| 237 | o3-mini-2025-01-31_mediumOpenAI | closed | effort medium | 2.1% | not dated | via Epoch AI |
| 238 | o3-pro-2025-06-10_lowOpenAI | closed | effort low | 2.1% | not dated | via Epoch AI |
| 239 | o3-2025-04-16_lowOpenAI | closed | effort low | 2.0% | not dated | via Epoch AI |
| 240 | gemini-2.5-flash-preview-05-20_16KGoogle DeepMind | closed | default | 2.0% | not dated | via Epoch AI |
| 241 | gpt-5-2025-08-07_lowOpenAI | closed | effort low | 1.9% | not dated | via Epoch AI |
| 242 | gpt-5-2025-08-07_lowOpenAI | closed | effort low | 1.9% | not dated | via Epoch AI |
| 243 | gpt-5.1-2025-11-13_lowOpenAI | closed | effort low | 1.9% | not dated | via Epoch AI |
| 244 | gpt-5.4-nano-2026-03-17_mediumOpenAI | closed | effort medium | 1.9% | not dated | via Epoch AI |
| 245 | o3-pro-2025-06-10_mediumOpenAI | closed | effort medium | 1.9% | not dated | via Epoch AI |
| 246 | gemini-2.5-flash-preview-05-20Google DeepMind | closed | default | 1.7% | not dated | via Epoch AI |
| 247 | claude-haiku-4-5-20251001_8KAnthropic | closed | default | 1.7% | not dated | via Epoch AI |
| 248 | gpt-5-mini-2025-08-07_minimalOpenAI | closed | effort minimal | 1.7% | not dated | via Epoch AI |
| 249 | o4-mini-2025-04-16_lowOpenAI | closed | effort low | 1.7% | not dated | via Epoch AI |
| 250 | gpt-6-sol_noneOpenAI | closed | effort none | 1.7% | not dated | via Epoch AI |
| 251 | gpt-5.4-nano-2026-03-17_lowOpenAI | closed | effort low | 1.5% | not dated | via Epoch AI |
| 252 | gemini-3.5-flash-lite_lowGoogle DeepMind | closed | effort low | 1.5% | not dated | via Epoch AI |
| 253 | Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_none, Alibaba | open | effort none | 1.5% | not dated | via Epoch AI |
| 254 | deepseek-ai/DeepSeek-R1Epoch: DeepSeek-R1, DeepSeek | open | default | 1.3% | not dated | via Epoch AI |
| 255 | gemini-2.0-flash-001Google DeepMind,Google | closed | default | 1.3% | not dated | via Epoch AI |
| 256 | claude-opus-4-20250514Anthropic | closed | default | 1.3% | not dated | via Epoch AI |
| 257 | claude-sonnet-4-20250514Anthropic | closed | default | 1.3% | not dated | via Epoch AI |
| 258 | codex-mini-2025-05-16 | not stated | default | 1.3% | not dated | via Epoch AI |
| 259 | claude-haiku-4-5-20251001Anthropic | closed | default | 1.3% | not dated | via Epoch AI |
| 260 | claude-haiku-4-5-20251001_1KAnthropic | closed | default | 1.3% | not dated | via Epoch AI |
| 261 | deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_none, DeepSeek | open | effort none | 1.3% | not dated | via Epoch AI |
| 262 | gemini-3-flash-previewGoogle DeepMind | closed | default | 1.3% | not dated | via Epoch AI |
| 263 | Qwen/Qwen3-235B-A22B-Instruct-2507Epoch: Qwen3-235B-A22B-Instruct-2507, Alibaba | open | default | 1.3% | not dated | via Epoch AI |
| 264 | deepseek-ai/DeepSeek-R1-0528Epoch: DeepSeek-R1-0528, DeepSeek | open | default | 1.1% | not dated | via Epoch AI |
| 265 | gpt-5.4-mini-2026-03-17_lowOpenAI | closed | effort low | 1.1% | not dated | via Epoch AI |
| 266 | claude-3-7-sonnet-20250219_8KAnthropic | closed | default | 0.9% | not dated | via Epoch AI |
| 267 | gpt-5-nano-2025-08-07_mediumOpenAI | closed | effort medium | 0.9% | not dated | via Epoch AI |
| 268 | claude-sonnet-4-20250514_1KAnthropic | closed | default | 0.9% | not dated | via Epoch AI |
| 269 | deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_none, DeepSeek | open | effort none | 0.8% | not dated | via Epoch AI |
| 270 | gemini-3.5-flash-lite_minimalGoogle DeepMind | closed | effort minimal | 0.8% | not dated | via Epoch AI |
| 271 | gpt-5-mini-2025-08-07_lowOpenAI | closed | effort low | 0.8% | not dated | via Epoch AI |
| 272 | gpt-5.2-2025-12-11_unknownOpenAI | closed | effort unknown | 0.8% | not dated | via Epoch AI |
| 273 | o1-mini-2024-09-12_unknownOpenAI | closed | effort unknown | 0.8% | not dated | via Epoch AI |
| 274 | gemini-1.5-pro-002Google DeepMind | closed | default | 0.8% | not dated | via Epoch AI |
| 275 | gpt-4.5-preview-2025-02-27OpenAI | closed | default | 0.8% | not dated | via Epoch AI |
| 276 | claude-3-7-sonnet-20250219_16KAnthropic | closed | default | 0.7% | not dated | via Epoch AI |
| 277 | gpt-4.1-2025-04-14OpenAI | closed | default | 0.4% | not dated | via Epoch AI |
| 278 | gpt-5.1-2025-11-13_noneOpenAI | closed | effort none | 0.4% | not dated | via Epoch AI |
| 279 | grok-3-mini_lowxAI | closed | effort low | 0.4% | not dated | via Epoch AI |
| 280 | claude-3-7-sonnet-20250219_1KAnthropic | closed | default | 0.4% | not dated | via Epoch AI |
| 281 | claude-3-7-sonnet-20250219Anthropic | closed | default | 0.0% | not dated | via Epoch AI |
| 282 | claude-opus-4-20250514_1KAnthropic | closed | default | 0.0% | not dated | via Epoch AI |
| 283 | gemini-2.5-pro-preview-06-05_1KGoogle DeepMind | closed | default | 0.0% | not dated | via Epoch AI |
| 284 | gpt-4.1-mini-2025-04-14OpenAI | closed | default | 0.0% | not dated | via Epoch AI |
| 285 | gpt-4.1-nano-2025-04-14OpenAI | closed | default | 0.0% | not dated | via Epoch AI |
| 286 | gpt-4o-2024-11-20OpenAI | closed | default | 0.0% | not dated | via Epoch AI |
| 287 | gpt-4o-mini-2024-07-18OpenAI | closed | default | 0.0% | not dated | via Epoch AI |
| 288 | gpt-5-2025-08-07_minimalOpenAI | closed | effort minimal | 0.0% | not dated | via Epoch AI |
| 289 | gpt-5-nano-2025-08-07_lowOpenAI | closed | effort low | 0.0% | not dated | via Epoch AI |
| 290 | gpt-5-nano-2025-08-07_minimalOpenAI | closed | effort minimal | 0.0% | not dated | via Epoch AI |
| 291 | gpt-6-luna_noneOpenAI | closed | effort none | 0.0% | not dated | via Epoch AI |
| 292 | grok-3xAI | closed | default | 0.0% | not dated | via Epoch AI |
| 293 | thinkingmachines/Inkling-SmallEpoch: inkling-small_none, Thinking Machines | open | effort none | 0.0% | not dated | via Epoch AI |
| 294 | meta-llama/Llama-4-Maverick-17B-128E-InstructEpoch: Llama-4-Maverick-17B-128E-Instruct, Meta AI | open | default | 0.0% | not dated | via Epoch AI |
| 295 | meta-llama/Llama-4-Scout-17B-16E-InstructEpoch: Llama-4-Scout-17B-16E-Instruct, Meta AI | open | default | 0.0% | not dated | via Epoch AI |
| 296 | magistral-medium-2506Mistral AI | closed | default | 0.0% | not dated | via Epoch AI |
| 297 | magistral-medium-2506Mistral AI | closed | default | 0.0% | not dated | via Epoch AI |
| 298 | mistralai/Magistral-Small-2506Epoch: magistral-small-2506, Mistral AI | open | default | 0.0% | not dated | via Epoch AI |
| 299 | o3-mini-2025-01-31_lowOpenAI | closed | effort low | 0.0% | not dated | via Epoch AI |
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