Benchmarks
All AI models: full benchmark table
All 271 models from the AI benchmarks leaderboard in one table, highest overall score first. The leaderboard has the tabs, search, filters, charts and model details; this page is the plain version.
Up to date. Data updated 1 of 1 source current Data status for all sources
Scores are percent correct; a dash (–) means “not tested”, not zero. The range after the overall score is Epoch's 90% confidence interval.
Rating key
The Rating column is our quick guide, worked out from Epoch AI's overall score (ECI); it is not a rating Epoch AI publishes. The cut-offs are percentiles of all 271 models' scores, so they move when the data updates.
- Overall score (ECI) 148.88 or higher: the top 25% (68 models).
- ECI from 118.48 up to 148.88: the middle 50% (135 models).
- ECI below 118.48: the bottom 25% (68 models).
| # | Rating | Model | Company | Released | Overall (ECI) | Science & facts score | GPQA Diamond | SimpleQA Verified | Competition math score | MATH Level 5 | OTIS Mock AIME | Research math score | FrontierMath T1–3 | FrontierMath Tier 4 | SWE-bench Verified | Puzzles score | Chess Puzzles | Mystery Game Puzzles |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | GPT-6 Astra | OpenAI | 169.1[166–174] | 85.7 | 95.8 | 75.6 | 100.0 | 95.7 | 93.7 | 97.6 | 80.0 | 76.0 | 84.0 | |||||
| 2 | Claude Opus 5.5 | Anthropic | 168.4[165–173] | 81.4 | 90.6 | 72.2 | 100.0 | 93.1 | 91.2 | 95.0 | 69.0 | 67.0 | 71.0 | |||||
| 3 | GPT-6.1 Terra | OpenAI | 168.1[165–173] | 75.8 | 94.3 | 57.3 | 100.0 | 91.8 | 90.9 | 92.7 | 74.0 | |||||||
| 4 | GPT-6.1 Sol | OpenAI | 167.8[164–173] | 84.7 | 95.4 | 73.9 | 100.0 | 96.8 | 93.7 | 100.0 | 74.0 | 68.0 | 80.0 | |||||
| 5 | Claude Sonnet 5.5 | Anthropic | 166.7[164–171] | 71.0 | 95.6 | 46.5 | 100.0 | 84.7 | 88.8 | 80.5 | 62.0 | 59.0 | 65.0 | |||||
| 6 | Claude Fable 5.1 | Anthropic | 166.4[164–171] | 70.8 | 100.0 | 89.0 | 90.2 | 87.8 | 53.0 | 48.0 | 58.0 | |||||||
| 7 | Gemini 4 Argon | Google DeepMind | 166.1[163–170] | 86.2 | 96.4 | 76.0 | 100.0 | 82.0 | 86.0 | 78.0 | 61.5 | 59.0 | 64.0 | |||||
| 8 | Claude Opus 5 | Anthropic | 164.5[162–168] | 76.9 | 93.9 | 59.9 | 98.9 | 79.4 | 85.6 | 73.2 | 51.0 | 43.0 | 59.0 | |||||
| 9 | GPT-6 Sol | OpenAI | 163.8[161–167] | 77.5 | 94.3 | 60.7 | 100.0 | 89.9 | 89.8 | 90.0 | 56.0 | 56.0 | 56.0 | |||||
| 10 | GPT-5.5 Pro | OpenAI | 163.1[160–167] | 93.9 | 100.0 | 82.8 | 87.7 | 78.0 | 62.0 | |||||||||
| 11 | Claude Fable 5 | Anthropic | 163.0[161–167] | 78.3 | 85.9 | 70.7 | 100.0 | 88.6 | 87.0 | 90.2 | 49.5 | 47.0 | 52.0 | |||||
| 12 | GPT-5.6 Sol | OpenAI | 162.6[160–166] | 81.6 | 93.5 | 69.7 | 100.0 | 86.0 | 89.1 | 82.9 | 63.0 | 68.0 | 58.0 | |||||
| 13 | Muse Wonder | Meta AI | 161.7[159–165] | 99.4 | 76.4 | 82.1 | 70.7 | 51.0 | ||||||||||
| 14 | Claude Haiku 5.5 | Anthropic | 160.4[157–164] | 56.7 | 89.6 | 23.8 | 98.9 | 60.7 | 75.1 | 46.3 | 42.5 | 55.0 | 30.0 | |||||
| 15 | GPT-5.6 Terra | OpenAI | 160.1[158–163] | 68.3 | 93.3 | 43.2 | 99.7 | 78.3 | 86.0 | 70.7 | 41.0 | 47.0 | 35.0 | |||||
| 16 | GPT-5.4 Pro | OpenAI | 159.8[157–163] | 70.4 | 94.6 | 46.3 | 70.5 | 82.5 | 58.5 | 59.0 | ||||||||
| 17 | Claude Opus 4.8 | Anthropic | 159.6[157–162] | 72.0 | 91.0 | 53.0 | 98.3 | 68.0 | 80.0 | 56.1 | 44.0 | 52.0 | 36.0 | |||||
| 18 | GPT-5.5 | OpenAI | 159.4[157–163] | 78.5 | 94.0 | 63.0 | 100.0 | 78.9 | 85.3 | 72.5 | 80.6 | 59.5 | 63.0 | 56.0 | ||||
| 19 | DeepSeek V4.1 Flash 1017Open weights | DeepSeek | 158.8[157–162] | 67.8 | 93.2 | 42.5 | 99.4 | 58.8 | 71.2 | 46.3 | 47.5 | 46.0 | 49.0 | |||||
| 20 | Kimi K3Open weights | Moonshot | 158.3[156–161] | 71.8 | 93.1 | 50.6 | 97.2 | 55.6 | 72.2 | 39.0 | 31.5 | 37.0 | 26.0 | |||||
| 21 | Gemini 3.7 Flash | Google DeepMind | 158.2[156–161] | 82.0 | 94.8 | 69.2 | 97.2 | 54.1 | 71.6 | 36.6 | 51.5 | 66.0 | 37.0 | |||||
| 22 | Qwen 3.8 Max | Alibaba | 158.0[156–161] | 69.3 | 92.7 | 45.8 | 99.4 | 60.5 | 74.7 | 46.3 | 42.0 | 46.0 | 38.0 | |||||
| 23 | Gemini 3.8 Flash | Google DeepMind | 157.6[155–161] | 82.6 | 95.4 | 69.7 | 98.9 | 45.2 | 68.4 | 22.0 | 52.5 | 58.0 | 47.0 | |||||
| 24 | Muse Spark 1.3 | Meta AI | 157.4[155–160] | 99.2 | 60.4 | 74.4 | 46.3 | 31.0 | 37.0 | 25.0 | ||||||||
| 25 | GPT-5.4 | OpenAI | 157.1[155–159] | 69.2 | 93.3 | 45.1 | 97.8 | 63.8 | 78.6 | 49.0 | 76.9 | 42.5 | 48.0 | 37.0 | ||||
| 26 | Claude Sonnet 5 | Anthropic | 157.0[154–160] | 62.1 | 90.5 | 33.7 | 94.7 | 47.4 | 65.6 | 29.3 | 40.0 | 45.0 | 35.0 | |||||
| 27 | Grok 4.6 | xAI | 156.9[155–159] | 71.7 | 94.0 | 49.3 | 99.2 | 48.9 | 66.0 | 31.7 | 33.5 | 33.0 | 34.0 | |||||
| 28 | GPT-6 Luna | OpenAI | 156.8[154–159] | 66.0 | 90.5 | 41.4 | 98.9 | 67.5 | 78.9 | 56.1 | 21.5 | 36.0 | 7.0 | |||||
| 29 | Claude Opus 4.7 | Anthropic | 156.7[155–159] | 71.0 | 90.2 | 51.7 | 97.8 | 51.0 | 70.2 | 31.7 | 83.5 | 34.5 | 41.0 | 28.0 | ||||
| 30 | GPT-5.6 Luna | OpenAI | 156.7[154–159] | 66.3 | 91.6 | 41.0 | 98.3 | 71.5 | 82.1 | 61.0 | 32.0 | 43.0 | 21.0 | |||||
| 31 | GPT-5.2 Pro | OpenAI | 156.5[154–161] | 60.0 | 74.0 | 46.0 | 60.0 | |||||||||||
| 32 | Qwen3.8 Max (0902) | Alibaba | 156.2[154–159] | 69.8 | 92.3 | 47.3 | 100.0 | 49.8 | 65.6 | 34.1 | 37.0 | |||||||
| 33 | GLM-5.3Open weights | Z.ai (Zhipu AI) | 156.2[154–159] | 66.0 | 90.9 | 41.0 | 91.1 | 49.0 | 68.8 | 29.3 | 35.0 | 37.0 | 33.0 | |||||
| 34 | DeepSeek V4.1 FlashOpen weights | DeepSeek | 156.1[154–158] | 66.3 | 89.8 | 42.9 | 98.3 | 47.1 | 67.4 | 26.8 | 45.5 | 48.0 | 43.0 | |||||
| 35 | DeepSeek V4 Pro 0813Open weights | DeepSeek | 155.9[154–158] | 72.3 | 91.7 | 52.9 | 98.6 | 45.7 | 64.6 | 26.8 | 44.5 | 46.0 | 43.0 | |||||
| 36 | Claude Opus 4.6 | Anthropic | 155.8[154–158] | 68.8 | 90.5 | 47.0 | 94.4 | 46.4 | 66.0 | 26.8 | 78.7 | 23.0 | 21.0 | 25.0 | ||||
| 37 | Gemini 3.1 Pro | Google DeepMind | 155.2[153–158] | 84.0 | 94.4 | 73.5 | 95.6 | 43.2 | 59.6 | 26.8 | 75.6 | 42.5 | 51.0 | 34.0 | ||||
| 38 | Muse Spark 1.2 | Meta AI | 155.2[153–158] | 60.3 | 27.0 | |||||||||||||
| 39 | Muse Spark OpenOpen weights | Meta AI | 155.1[153–157] | 72.6 | 88.6 | 56.6 | 98.1 | 50.8 | 69.8 | 31.7 | ||||||||
| 40 | Gemini 3.5 Flash | Google DeepMind | 155.0[153–157] | 79.5 | 92.8 | 66.2 | 95.6 | 44.8 | 62.8 | 26.8 | 79.3 | 38.5 | 45.0 | 32.0 | ||||
| 41 | DeepSeek V4 Flash 0731Open weights | DeepSeek | 154.9[153–157] | 62.3 | 91.0 | 33.6 | 94.4 | 41.0 | 57.5 | 24.4 | 33.0 | 32.0 | 34.0 | |||||
| 42 | Gemini 3.6 Flash | Google DeepMind | 154.7[153–157] | 80.2 | 94.1 | 66.2 | 94.2 | 40.5 | 58.9 | 22.0 | 36.0 | 42.0 | 30.0 | |||||
| 43 | Grok 4.7 | xAI | 154.6[153–157] | 74.3 | 92.7 | 56.0 | 98.1 | 35.0 | 53.0 | 17.1 | 35.0 | 41.0 | 29.0 | |||||
| 44 | Grok 4.5 | xAI | 154.4[153–157] | 70.8 | 93.4 | 48.3 | 97.8 | 40.8 | 57.2 | 24.4 | 35.0 | |||||||
| 45 | Qwen3.7-Max | Alibaba | 154.2[152–157] | 73.3 | 90.9 | 55.8 | 95.6 | 49.3 | 64.6 | 34.1 | 77.3 | 29.0 | 26.0 | 32.0 | ||||
| 46 | Muse Spark 1.1 | Meta AI | 154.2[151–156] | 57.8 | 34.0 | |||||||||||||
| 47 | GPT-5.2 | OpenAI | 153.8[152–156] | 64.3 | 91.4 | 37.1 | 96.1 | 49.6 | 67.4 | 31.7 | 73.8 | 35.5 | 48.0 | 23.0 | ||||
| 48 | GLM-5.3-FlashOpen weights | Z.ai (Zhipu AI) | 153.2[151–155] | 90.2 | 93.9 | 36.5 | 55.8 | 17.1 | 14.5 | 21.0 | 8.0 | |||||||
| 49 | Gemini 3 Pro | Google DeepMind | 153.2[151–156] | 92.6 | 91.4 | 72.9 | ||||||||||||
| 50 | Claude Sonnet 4.6 | Anthropic | 152.9[150–155] | 61.5 | 87.4 | 35.5 | 85.8 | 75.2 | 19.5 | 23.0 | 16.0 | |||||||
| 51 | Grok 4.20 | xAI | 152.3[150–155] | 59.8 | 89.3 | 30.2 | 92.2 | 31.0 | 44.9 | 17.1 | 29.0 | |||||||
| 52 | Muse Spark | Meta AI | 152.2[150–154] | 89.8 | 88.9 | |||||||||||||
| 53 | GLM-5.2Open weights | Z.ai (Zhipu AI) | 152.1[150–154] | 63.1 | 91.9 | 34.2 | 86.4 | 44.3 | 59.2 | 29.3 | 78.7 | 20.5 | 22.0 | 19.0 | ||||
| 54 | Gemini 3 Flash | Google DeepMind | 152.0[151–154] | 78.1 | 89.4 | 66.8 | 95.6 | 34.2 | 51.2 | 17.1 | 75.4 | 32.5 | 39.0 | 26.0 | ||||
| 55 | Qwen 3.8 27BOpen weights | Alibaba | 151.8[150–154] | |||||||||||||||
| 56 | Kimi K2.6Open weights | Moonshot | 151.1[149–153] | 62.8 | 90.8 | 34.9 | 96.1 | 41.4 | 57.2 | 25.6 | 76.7 | 19.0 | 20.0 | 18.0 | ||||
| 57 | GPT-5 Pro | OpenAI | 150.7[149–155] | 37.6 | 55.8 | 19.5 | 48.0 | |||||||||||
| 58 | Mistral Large 4Open weights | Mistral AI | 150.6[147–153] | 56.3 | 90.2 | 22.4 | 99.4 | 31.7 | 46.3 | 17.1 | 24.5 | 29.0 | 20.0 | |||||
| 59 | Inkling-SmallOpen weights | Thinking Machines | 150.5[148–152] | 53.8 | 88.5 | 19.1 | 90.0 | 31.7 | 46.3 | 17.1 | 6.0 | |||||||
| 60 | Beam 1.0Open weights | Reflection | 150.2[149–152] | |||||||||||||||
| 61 | Kimi K2.7 CodeOpen weights | Moonshot | 150.2[148–152] | 62.2 | 87.9 | 36.5 | 95.6 | 33.1 | 54.0 | 12.2 | 14.0 | |||||||
| 62 | Claude Opus 4.5 | Anthropic | 150.2[148–153] | 65.8 | 86.0 | 45.7 | 86.1 | 19.6 | 34.4 | 4.9 | 76.7 | 18.0 | 14.0 | 22.0 | ||||
| 63 | GPT-5 | OpenAI | 150.0 | 68.2 | 86.2 | 50.1 | 94.8 | 98.1 | 91.4 | 38.7 | 55.4 | 22.0 | 73.6 | 29.5 | 36.0 | 23.0 | ||
| 64 | GPT-5.1 | OpenAI | 149.9[149–152] | 67.8 | 87.6 | 48.0 | 88.6 | 68.0 | 30.5 | 42.0 | 19.0 | |||||||
| 65 | GLM-5.1Open weights | Z.ai (Zhipu AI) | 149.9[148–152] | 62.0 | 89.9 | 34.0 | 93.3 | 36.8 | 74.2 | 20.0 | ||||||||
| 66 | Qwen 3.6 Max (Preview) | Alibaba | 149.4[148–153] | 69.7 | 87.4 | 52.0 | 91.1 | 76.7 | 19.0 | |||||||||
| 67 | DeepSeek-V4-ProOpen weights | DeepSeek | 149.1[147–151] | 69.0 | 90.9 | 47.0 | 96.7 | 23.8 | 45.3 | 2.4 | 77.6 | 18.0 | 19.0 | 17.0 | ||||
| 68 | Grok 4.3 Beta | xAI | 148.9[147–150] | 61.0 | 88.8 | 33.2 | 93.3 | 28.7 | 42.8 | 14.6 | 18.0 | |||||||
| 69 | GPT-5.4 Mini | OpenAI | 148.8[147–151] | 58.2 | 86.9 | 29.4 | 88.9 | 30.5 | 51.2 | 9.8 | 17.0 | 23.0 | 11.0 | |||||
| 70 | InklingOpen weights | Thinking Machines | 148.5[146–150] | 64.3 | 88.3 | 40.3 | 88.9 | 19.1 | 33.3 | 4.9 | 21.0 | |||||||
| 71 | Kimi K2.5Open weights | Moonshot | 148.1[146–149] | 60.9 | 87.6 | 34.3 | 92.2 | 73.8 | ||||||||||
| 72 | Qwen 3.6 Plus | Alibaba | 147.8[146–150] | 66.3 | 88.4 | 44.1 | 93.3 | 38.2 | 57.9 | 13.0 | 14.0 | 12.0 | ||||||
| 73 | o3-pro | OpenAI | 147.4[146–150] | |||||||||||||||
| 74 | Qwen3.7-Plus | Alibaba | 147.1[145–149] | 87.9 | 93.3 | 34.4 | 17.0 | |||||||||||
| 75 | MiniMax-M3Open weights | MiniMax | 146.9[142–150] | 90.9 | 71.1 | 10.5 | 13.0 | 8.0 | ||||||||||
| 76 | o3 | OpenAI | 146.9[145–149] | 65.6 | 81.8 | 49.4 | 91.1 | 97.8 | 84.4 | 33.3 | 62.3 | 33.0 | 37.0 | 29.0 | ||||
| 77 | Qwen3.5 397B-A17BOpen weights | Alibaba | 146.8[145–149] | 86.4 | 88.9 | 31.2 | 18.0 | |||||||||||
| 78 | Claude Sonnet 4.5 | Anthropic | 146.8[145–149] | 56.5 | 82.3 | 30.7 | 87.8 | 97.7 | 77.8 | 13.1 | 23.9 | 2.4 | 71.3 | 13.5 | 10.0 | 17.0 | ||
| 79 | MiniMax-M2.5Open weights | MiniMax | 146.7[141–148] | |||||||||||||||
| 80 | Grok 4 | xAI | 146.3[144–148] | 87.0 | 84.0 | |||||||||||||
| 81 | Qwen3.6 27BOpen weights | Alibaba | 146.3[144–148] | 85.9 | 91.1 | 35.1 | 7.0 | |||||||||||
| 82 | Qwen 3.5 Plus (hosted 397B-A17B) | Alibaba | 146.3[144–148] | 55.1 | 84.8 | 25.4 | 86.7 | 12.5 | 8.0 | 17.0 | ||||||||
| 83 | DeepSeek-V4-FlashOpen weights | DeepSeek | 146.2[144–148] | |||||||||||||||
| 84 | GPT-5.4 Nano | OpenAI | 146.1[144–148] | 45.1 | 78.5 | 11.7 | 87.8 | 28.5 | 44.9 | 12.2 | 27.0 | 45.0 | 9.0 | |||||
| 85 | DeepSeek-V3.2Open weights | DeepSeek | 146.1[144–147] | 83.4 | 87.8 | |||||||||||||
| 86 | Nemotron 3 UltraOpen weights | Nvidia | 146.0[144–148] | 85.4 | 86.7 | 20.0 | ||||||||||||
| 87 | MiniMax-M2.7Open weights | MiniMax | 145.9[139–148] | |||||||||||||||
| 88 | Kimi K2 ThinkingOpen weights | Moonshot | 145.9[143–147] | 84.2 | 83.1 | |||||||||||||
| 89 | GLM-5Open weights | Z.ai (Zhipu AI) | 145.8[144–148] | 87.8 | 80.0 | 72.1 | ||||||||||||
| 90 | o4-mini | OpenAI | 145.6[143–147] | 49.6 | 79.6 | 19.6 | 89.8 | 97.8 | 81.7 | 20.5 | 36.1 | 4.9 | 13.5 | 22.0 | 5.0 | |||
| 91 | GPT-5 mini | OpenAI | 145.4[144–147] | 48.3 | 75.0 | 21.6 | 92.3 | 97.8 | 86.7 | 29.5 | 46.7 | 12.2 | 64.7 | 17.0 | 24.0 | 10.0 | ||
| 92 | Gemini 2.5 Pro (Jun 2025) | Google DeepMind | 145.3[144–147] | 85.3 | 84.7 | 12.3 | 24.6 | 0.0 | 57.6 | |||||||||
| 93 | DeepSeek-V3.2-ExpOpen weights | DeepSeek | 145.0[143–147] | |||||||||||||||
| 94 | Gemini 3.5 Flash-Lite | Google DeepMind | 145.0[143–146] | 83.3 | 71.1 | 13.0 | 26.0 | 0.0 | 20.0 | 21.0 | 19.0 | |||||||
| 95 | Gemini 3.1 Flash-Lite | 144.3[142–146] | 81.8 | 80.0 | 27.7 | |||||||||||||
| 96 | Grok 4 Fast | xAI | 144.2[142–146] | |||||||||||||||
| 97 | Qwen3.7 Flash | Alibaba | 144.1[142–146] | 82.3 | 86.7 | 19.3 | 15.0 | |||||||||||
| 98 | Gemini 2.5 Pro (Mar 2025) | Google DeepMind | 144.0[142–147] | 83.8 | 95.6 | |||||||||||||
| 99 | Claude Opus 4.1 | Anthropic | 143.9[142–146] | 77.3 | 68.9 | 7.5 | 12.6 | 2.4 | 73.3 | 21.0 | ||||||||
| 100 | Qwen 3.6 35B-A3BOpen weights | Alibaba | 143.8[141–146] | 84.8 | 86.7 | 20.4 | 22.0 | |||||||||||
| 101 | Qwen3-235B-A22B-Thinking (Jul 2025)Open weights | Alibaba | 143.8[142–145] | 60.3 | 80.1 | 40.4 | 86.7 | 9.0 | ||||||||||
| 102 | GLM-4.7Open weights | Z.ai (Zhipu AI) | 143.7[141–146] | 57.8 | 83.3 | 32.2 | 83.3 | 21.0 | ||||||||||
| 103 | Qwen 3.5 Flash (hosted 35B-A3B) | Alibaba | 143.6[141–145] | 51.3 | 82.3 | 20.3 | 84.4 | 18.2 | 17.0 | 14.0 | 20.0 | |||||||
| 104 | Gemma 4 31B ITOpen weights | Google DeepMind | 143.1[141–146] | 43.1 | 75.8 | 10.4 | 73.3 | |||||||||||
| 105 | Qwen 3.6 Flash | Alibaba | 143.0[140–144] | 49.6 | 83.3 | 15.9 | 84.4 | 22.5 | 15.0 | 12.0 | 18.0 | |||||||
| 106 | Gemini 2.5 Flash (Sep 2025) | Google DeepMind | 142.9[139–146] | |||||||||||||||
| 107 | Qwen3.5-35B-A3BOpen weights | Alibaba | 142.6[140–145] | 83.5 | 70.0 | |||||||||||||
| 108 | GPT-5.5 Instant | OpenAI | 142.5[140–145] | 82.5 | 68.1 | 14.3 | 26.3 | 2.4 | ||||||||||
| 109 | Claude Opus 4 | Anthropic | 142.4[140–144] | 76.3 | 74.7 | 85.0 | 64.4 | 70.7 | ||||||||||
| 110 | Qwen3-Max | Alibaba | 142.4[140–145] | 60.6 | 72.6 | 48.7 | 85.2 | 97.1 | 73.3 | 18.9 | 5.0 | |||||||
| 111 | Claude Haiku 4.5 | Anthropic | 142.2[139–144] | 42.2 | 71.2 | 13.2 | 81.6 | 96.4 | 66.7 | 8.0 | ||||||||
| 112 | Gemini 2.5 Pro (May 2025) | Google DeepMind | 142.2[138–145] | 66.7 | 95.9 | |||||||||||||
| 113 | Gemma 4 26B A4BOpen weights | Google DeepMind | 141.9[139–144] | 73.2 | 82.2 | |||||||||||||
| 114 | o1 | OpenAI | 141.8[140–143] | 59.0 | 76.8 | 41.1 | 84.0 | 94.7 | 73.3 | 14.7 | 10.0 | |||||||
| 115 | Claude Sonnet 4 | Anthropic | 141.6[139–143] | 79.2 | 77.8 | 84.4 | 71.1 | |||||||||||
| 116 | Gemini 2.5 Flash (May 2025) | Google DeepMind | 141.5[139–143] | 70.8 | ||||||||||||||
| 117 | Mistral Medium 3.5Open weights | Mistral AI | 141.5[139–144] | |||||||||||||||
| 118 | DeepSeek-R1 (May 2025)Open weights | DeepSeek | 141.2[139–143] | 76.3 | 81.5 | 96.6 | 66.4 | |||||||||||
| 119 | Claude 3.7 Sonnet | Anthropic | 141.0[139–143] | 79.7 | 74.5 | 91.2 | 57.8 | 61.0 | ||||||||||
| 120 | Gemini 2.5 Flash (Jun 2025) | Google DeepMind | 140.6[138–142] | |||||||||||||||
| 121 | o3-mini | OpenAI | 140.3[138–142] | 46.1 | 77.0 | 15.3 | 86.7 | 96.5 | 76.9 | 9.3 | 18.6 | 0.0 | 11.0 | 15.0 | 7.0 | |||
| 122 | Grok-3 mini | xAI | 140.3[138–142] | 76.3 | 84.3 | 90.9 | 77.8 | |||||||||||
| 123 | Kimi K2 (Jul 2025)Open weights | Moonshot | 139.9[136–142] | |||||||||||||||
| 124 | Gemini 2.5 Flash (Apr 2025) | Google DeepMind | 139.9[136–142] | 73.1 | ||||||||||||||
| 125 | gpt-oss-120bOpen weights | OpenAI | 139.8[136–142] | 75.8 | 88.9 | 2.0 | ||||||||||||
| 126 | DeepSeek-V3.1Open weights | DeepSeek | 139.8[136–143] | |||||||||||||||
| 127 | Qwen3-30B-A3B-Thinking (Jul 2025)Open weights | Alibaba | 139.6[136–141] | 70.1 | 70.3 | |||||||||||||
| 128 | Qwen3.5-9BOpen weights | Alibaba | 139.4[136–141] | 79.0 | 61.7 | |||||||||||||
| 129 | GPT-5 nano | OpenAI | 139.3[134–141] | 40.6 | 69.4 | 11.7 | 88.2 | 95.2 | 81.1 | 11.2 | 20.0 | 2.4 | 15.5 | 22.0 | 9.0 | |||
| 130 | Qwen3-235B-A22BOpen weights | Alibaba | 139.1[135–141] | 70.7 | 68.9 | |||||||||||||
| 131 | DeepSeek-R1Open weights | DeepSeek | 138.8[137–140] | 71.7 | 73.2 | 93.1 | 53.3 | |||||||||||
| 132 | Qwen3-235B-A22B-Instruct (Jul 2025)Open weights | Alibaba | 138.8[136–141] | |||||||||||||||
| 133 | Qwen3-32BOpen weights | Alibaba | 138.5[135–140] | 65.7 | 66.9 | |||||||||||||
| 134 | Grok 3 | xAI | 138.3[136–140] | 75.8 | 72.2 | 88.7 | 55.6 | |||||||||||
| 135 | Qwen3-14BOpen weights | Alibaba | 138.2[134–140] | 63.8 | 66.4 | |||||||||||||
| 136 | gpt-oss-20bOpen weights | OpenAI | 137.8[133–140] | 60.8 | 65.3 | |||||||||||||
| 137 | QwQ-32BOpen weights | Alibaba | 137.5[133–141] | 65.3 | 59.2 | |||||||||||||
| 138 | Qwen3-30B-A3B-Instruct (Jul 2025)Open weights | Alibaba | 137.3[132–139] | 55.6 | 62.2 | |||||||||||||
| 139 | GPT-4.1 | OpenAI | 136.7[134–138] | 49.0 | 66.9 | 31.1 | 60.6 | 83.0 | 38.3 | 6.0 | 48.5 | 3.0 | ||||||
| 140 | GPT-4.5 | OpenAI | 136.7[134–138] | 68.7 | 58.2 | 78.6 | 37.8 | |||||||||||
| 141 | Qwen3-30B-A3BOpen weights | Alibaba | 136.2[131–139] | 61.7 | 62.8 | |||||||||||||
| 142 | Qwen3-8BOpen weights | Alibaba | 136.1[130–138] | 56.8 | 56.1 | |||||||||||||
| 143 | DeepSeek-V3 (Mar 2025)Open weights | DeepSeek | 135.9[132–138] | 67.6 | 56.6 | 75.5 | 37.8 | |||||||||||
| 144 | o1-mini | OpenAI | 135.8[132–137] | 62.4 | 68.0 | 89.2 | 46.9 | |||||||||||
| 145 | Gemini 2.0 Flash Thinking (Jan 2025) | Google DeepMind,Google | 135.3[130–138] | 57.1 | 57.8 | |||||||||||||
| 146 | Gemini 2.0 Pro | Google DeepMind | 135.0[131–137] | 65.7 | 83.5 | |||||||||||||
| 147 | GPT-4.1 mini | OpenAI | 134.9[131–136] | 39.3 | 65.8 | 12.7 | 66.0 | 87.3 | 44.7 | 6.7 | 4.0 | 1.0 | 7.0 | |||||
| 148 | Gemini 2.0 Flash (Feb 2025) | Google DeepMind,Google | 134.7[131–136] | 64.1 | 56.7 | 82.2 | 31.1 | |||||||||||
| 149 | o1-preview | OpenAI | 134.6[131–138] | 50.3 | 56.3 | 81.6 | 31.1 | |||||||||||
| 150 | Mistral Medium 3 | Mistral AI | 134.0[130–135] | 59.5 | 56.9 | 81.6 | 32.2 | |||||||||||
| 151 | Gemini 2.5 Flash-Lite (Jun 2025) | Google DeepMind | 133.8[130–136] | |||||||||||||||
| 152 | Claude 3.5 Sonnet (October 2024) | Anthropic | 133.4[128–138] | 55.3 | 32.7 | 56.9 | 8.5 | |||||||||||
| 153 | Magistral Small 1.0Open weights | Mistral AI | 133.1[129–134] | 56.1 | 30.0 | |||||||||||||
| 154 | Qwen2.5-Max | Alibaba | 132.5[129–136] | 56.1 | 41.7 | 67.2 | 16.1 | |||||||||||
| 155 | Llama 4 MaverickOpen weights | Meta AI | 132.3[129–134] | 67.0 | 46.8 | 73.0 | 20.6 | |||||||||||
| 156 | DeepSeek-V3Open weights | DeepSeek | 132.3[128–136] | 56.5 | 40.4 | 64.9 | 15.8 | |||||||||||
| 157 | Gemini 1.5 Pro (Sept 2024) | Google DeepMind | 131.7[128–133] | 57.2 | 46.8 | 70.4 | 23.1 | |||||||||||
| 158 | Phi-4Open weights | Microsoft Research | 130.6[125–132] | 56.1 | 39.4 | 64.9 | 13.8 | |||||||||||
| 159 | Grok-2 (Dec 2024) | xAI | 130.6[127–132] | 53.8 | 37.5 | 63.5 | 11.5 | |||||||||||
| 160 | Claude 3.5 Sonnet | Anthropic | 130.0 | 54.0 | 29.1 | 51.7 | 6.5 | |||||||||||
| 161 | Gemma 3 27BOpen weights | Google DeepMind | 129.9[124–132] | 47.7 | 48.3 | 74.0 | 22.5 | |||||||||||
| 162 | Llama 4 ScoutOpen weights | Meta AI | 129.6[125–131] | 51.8 | 35.0 | 62.3 | 7.8 | |||||||||||
| 163 | GPT-4.1 nano | OpenAI | 129.6[124–132] | 27.4 | 48.9 | 6.0 | 49.5 | 70.0 | 28.9 | 0.0 | ||||||||
| 164 | Gemini 1.5 Flash (Sep 2024) | Google DeepMind | 129.3[123–131] | 47.3 | 39.1 | 61.9 | 16.3 | |||||||||||
| 165 | GPT-4o (May 2024) | OpenAI | 129.0[125–131] | 48.9 | 28.6 | 51.0 | 6.3 | |||||||||||
| 166 | Qwen2.5-72BOpen weights | Alibaba | 128.9[124–131] | 49.1 | 35.6 | 63.2 | 8.1 | |||||||||||
| 167 | GPT-4o (Aug 2024) | OpenAI | 128.8[125–131] | 37.6 | 49.2 | 26.0 | 29.8 | 53.3 | 6.4 | 0.4 | 6.0 | |||||||
| 168 | Llama 3.1-405BOpen weights | Meta AI | 128.8[124–131] | 50.9 | 29.8 | 49.8 | 9.7 | |||||||||||
| 169 | GPT-4o (Nov 2024) | OpenAI | 128.7[124–131] | 47.9 | 28.0 | 49.8 | 6.3 | 31.0 | ||||||||||
| 170 | Mistral Large 2 (Nov 2024)Open weights | Mistral AI | 128.6[124–131] | 51.3 | 29.0 | 50.3 | 7.8 | |||||||||||
| 171 | Mistral Large 2 (Jul 2024)Open weights | Mistral AI | 127.6[123–129] | 49.0 | 26.6 | 44.8 | 8.5 | |||||||||||
| 172 | Mistral Small 3.1Open weights | Mistral AI | 127.5[123–129] | 47.5 | 26.3 | 46.8 | 5.8 | |||||||||||
| 173 | Llama 3.3 70BOpen weights | Meta AI | 127.3[123–129] | 47.4 | 23.4 | 41.6 | 5.1 | |||||||||||
| 174 | GPT-4 Turbo (Apr 2024) | OpenAI | 127.3[122–129] | 46.6 | 26.7 | 46.7 | 6.7 | 0.7 | ||||||||||
| 175 | Claude 3.5 Haiku | Anthropic | 127.0[121–129] | 38.1 | 25.3 | 46.4 | 4.3 | |||||||||||
| 176 | Gemini 1.5 Pro (May 2024) | Google DeepMind | 126.9[122–129] | 45.9 | 23.8 | 40.7 | 6.8 | |||||||||||
| 177 | Claude 3 Opus | Anthropic | 126.9[122–129] | 29.9 | 47.2 | 12.6 | 21.1 | 37.5 | 4.7 | 3.0 | ||||||||
| 178 | GPT-4o mini | OpenAI | 126.5[121–129] | 23.0 | 37.7 | 8.3 | 29.8 | 52.6 | 6.9 | 0.7 | 7.5 | 3.0 | 12.0 | |||||
| 179 | GPT-4 Turbo (Nov 2023) | OpenAI | 126.4[122–129] | 42.4 | 40.0 | |||||||||||||
| 180 | Llama 3.1-70BOpen weights | Meta AI | 125.9[121–128] | 44.2 | 20.2 | 36.7 | 3.6 | |||||||||||
| 181 | GPT-4 (Mar 2023) | OpenAI | 125.7[121–130] | 35.7 | 0.6 | |||||||||||||
| 182 | Llama 3.2 90BOpen weights | Meta AI | 125.5[120–127] | 41.0 | 21.0 | 39.4 | 2.6 | |||||||||||
| 183 | Qwen2-72BOpen weights | Alibaba | 125.3[119–127] | 40.8 | 39.1 | |||||||||||||
| 184 | DeepSeek-V2 (MoE-236B, May 2024)Open weights | DeepSeek | 124.7[119–128] | |||||||||||||||
| 185 | Gemma 3 12BOpen weights | Google DeepMind | 123.5[117–128] | 39.5 | 16.7 | |||||||||||||
| 186 | GPT-4 (Jun 2023) | OpenAI | 123.0[117–127] | 30.7 | 12.1 | 23.0 | 1.1 | 12.0 | ||||||||||
| 187 | Llama 3-70BOpen weights | Meta AI | 123.0[117–126] | 40.6 | 13.5 | 22.6 | 4.3 | |||||||||||
| 188 | Gemini 1.5 Flash (May 2024) | Google DeepMind | 122.6[117–125] | 40.4 | 14.5 | 25.1 | 3.9 | |||||||||||
| 189 | Gemma 2 27BOpen weights | Google DeepMind | 122.1[116–124] | 36.5 | 14.6 | 27.9 | 1.4 | |||||||||||
| 190 | Mistral Large | Mistral AI | 122.0[116–125] | 38.8 | 13.2 | 24.5 | 1.9 | |||||||||||
| 191 | Mixtral 8x22BOpen weights | Mistral AI | 122.0[116–124] | 34.1 | 24.2 | |||||||||||||
| 192 | phi-3-small 7.4BOpen weights | Microsoft | 121.8[115–125] | |||||||||||||||
| 193 | phi-3-medium 14BOpen weights | Microsoft | 121.2[115–124] | 27.6 | 17.6 | |||||||||||||
| 194 | Claude 3 Sonnet | Anthropic | 120.8[114–124] | 40.6 | 10.3 | 18.2 | 2.5 | |||||||||||
| 195 | Claude Instant | Anthropic | 120.2[113–123] | |||||||||||||||
| 196 | Claude 2 | Anthropic | 120.1[114–124] | 34.7 | 7.1 | 11.7 | 2.5 | |||||||||||
| 197 | Gemma 2 9BOpen weights | Google DeepMind | 119.7[112–122] | 27.5 | 10.8 | 21.0 | 0.6 | |||||||||||
| 198 | Qwen2.5-Coder-32BOpen weights | Alibaba | 119.4[111–125] | |||||||||||||||
| 199 | Claude 2.1 | Anthropic | 119.2[113–122] | 33.0 | 1.9 | |||||||||||||
| 200 | Command R+Open weights | Cohere,Cohere Labs (formerly Cohere for AI) | 119.0[110–125] | |||||||||||||||
| 201 | Claude 3 Haiku | Anthropic | 118.8[112–121] | 36.3 | 8.3 | 14.9 | 1.8 | |||||||||||
| 202 | Mistral NeMoOpen weights | Mistral AI | 118.6[112–122] | 29.9 | 10.8 | |||||||||||||
| 203 | GPT-3.5 Turbo (Nov 2023) | OpenAI | 118.5[109–121] | 28.0 | 15.9 | |||||||||||||
| 204 | Qwen2.5-7BOpen weights | Alibaba | 118.5[111–121] | 35.5 | 2.5 | |||||||||||||
| 205 | Mixtral 8x7BOpen weights | Mistral AI | 118.4[111–121] | 30.6 | 10.0 | |||||||||||||
| 206 | Ministral 3B | Mistral AI | 118.0[106–122] | 25.3 | 14.4 | |||||||||||||
| 207 | Yi-34BOpen weights | 01.AI | 117.3[110–120] | 14.7 | 5.1 | |||||||||||||
| 208 | phi-3-mini 3.8BOpen weights | Microsoft | 117.3[109–121] | |||||||||||||||
| 209 | Gemini 1.0 Pro | Google DeepMind | 117.1[111–120] | 34.0 | 6.1 | 11.2 | 1.1 | |||||||||||
| 210 | Stable Beluga 2Open weights | Stability AI | 117.0[110–120] | |||||||||||||||
| 211 | Llama 3-8BOpen weights | Meta AI | 116.4[108–119] | 26.1 | 4.0 | 6.1 | 1.9 | |||||||||||
| 212 | Llama 3.1-8BOpen weights | Meta AI | 116.4[107–122] | 27.0 | 12.3 | 22.9 | 1.7 | |||||||||||
| 213 | Qwen2.5-Coder-14B | Unknown | 116.3[107–122] | |||||||||||||||
| 214 | Gemma 3 4BOpen weights | Google DeepMind | 115.8[98–124] | 23.2 | 7.5 | |||||||||||||
| 215 | GPT-3.5 Turbo (Jan 2024) | OpenAI | 115.6[108–119] | 27.2 | 6.9 | 11.6 | 2.2 | 0.0 | 3.0 | |||||||||
| 216 | PaLM 2-L | Unknown | 114.9[106–129] | |||||||||||||||
| 217 | Llama 2-70BOpen weights | Meta AI | 113.7[106–118] | 26.3 | 1.6 | 3.3 | 0.0 | |||||||||||
| 218 | GPT-3.5 Turbo (Jun 2023) | OpenAI | 113.3[105–118] | |||||||||||||||
| 219 | Qwen2.5-Coder (7B)Open weights | Alibaba | 113.1[103–119] | |||||||||||||||
| 220 | Qwen-14BOpen weights | Alibaba | 112.9[105–117] | |||||||||||||||
| 221 | Mistral 7B v0.1Open weights | Mistral AI | 112.1[104–116] | |||||||||||||||
| 222 | Falcon-180BOpen weights | Technology Innovation Institute | 112.0[104–118] | |||||||||||||||
| 223 | internlm-20b | Unknown | 112.0[104–117] | |||||||||||||||
| 224 | Gemma 7BOpen weights | Google DeepMind | 111.9[104–116] | |||||||||||||||
| 225 | LLaMA-65BOpen weights | Meta AI | 110.1[102–114] | |||||||||||||||
| 226 | Falcon 2 11BOpen weights | Technology Innovation Institute | 109.5[101–115] | |||||||||||||||
| 227 | Mistral 7B v0.3Open weights | Mistral AI | 109.0[98–113] | 15.2 | 2.0 | 3.7 | 0.3 | |||||||||||
| 228 | DeepSeek-Coder-V2-Lite-Base | Unknown | 108.8[99–114] | |||||||||||||||
| 229 | PaLM 2-M | Unknown | 108.2[99–119] | |||||||||||||||
| 230 | Phi-2Open weights | Microsoft | 107.9[89–114] | |||||||||||||||
| 231 | Qwen2.5-Coder-3B | Unknown | 107.7[94–115] | |||||||||||||||
| 232 | Nemotron-4 15B | Nvidia | 107.6[99–113] | |||||||||||||||
| 233 | Yi-9B | Unknown | 107.5[98–115] | |||||||||||||||
| 234 | LLaMA-33BOpen weights | Meta AI | 107.3[99–112] | |||||||||||||||
| 235 | Qwen-7BOpen weights | Alibaba | 106.7[97–112] | |||||||||||||||
| 236 | PaLM 2-S | Unknown | 106.1[96–114] | |||||||||||||||
| 237 | Llama 2-13BOpen weights | Meta AI | 105.7[96–111] | |||||||||||||||
| 238 | Llama 2-34B | Meta AI | 105.1[96–110] | |||||||||||||||
| 239 | StarCoder 2 15BOpen weights | Hugging Face,ServiceNow,Nvidia,BigCode | 104.9[93–112] | |||||||||||||||
| 240 | Yi 6BOpen weights | 01.AI | 104.7[95–110] | |||||||||||||||
| 241 | Falcon-40BOpen weights | Technology Innovation Institute | 104.3[96–110] | |||||||||||||||
| 242 | Baichuan2-13BOpen weights | Baichuan | 103.1[90–110] | |||||||||||||||
| 243 | Qwen2.5-Coder (1.5B)Open weights | Alibaba | 102.9[87–111] | |||||||||||||||
| 244 | internlm-7b | Unknown | 102.8[92–109] | |||||||||||||||
| 245 | INTELLECT-1 | Prime Intellect,Hugging Face,Arcee AI | 100.8[90–106] | |||||||||||||||
| 246 | MPT-30BOpen weights | MosaicML | 100.6[91–106] | |||||||||||||||
| 247 | LLaMA-13BOpen weights | Meta AI | 100.5[91–107] | |||||||||||||||
| 248 | Llama 2-7BOpen weights | Meta AI | 98.8[88–106] | |||||||||||||||
| 249 | chatglm2-6b | Unknown | 98.8[82–106] | |||||||||||||||
| 250 | LLaMA-7BOpen weights | Meta AI | 96.9[86–104] | |||||||||||||||
| 251 | Baichuan 2-7BOpen weights | Baichuan | 96.3[85–105] | |||||||||||||||
| 252 | DeepSeek Coder 33BOpen weights | DeepSeek,Peking University | 96.2[84–103] | |||||||||||||||
| 253 | Falcon-7BOpen weights | Technology Innovation Institute | 95.0[82–102] | |||||||||||||||
| 254 | CodeQwen1.5-7B | Unknown | 94.8[80–103] | |||||||||||||||
| 255 | vicuna-13b-v1.1 | Unknown | 94.5[82–102] | |||||||||||||||
| 256 | MPT-7BOpen weights | MosaicML | 94.5[81–102] | |||||||||||||||
| 257 | Gemma 2BOpen weights | Google DeepMind | 94.1[82–101] | |||||||||||||||
| 258 | StarCoder 2 7BOpen weights | Hugging Face,ServiceNow,Nvidia,BigCode | 93.5[78–101] | |||||||||||||||
| 259 | XGen-7BOpen weights | Salesforce | 93.3[82–101] | |||||||||||||||
| 260 | Qwen-1_8B | Unknown | 92.8[73–102] | |||||||||||||||
| 261 | open_llama_7b | Unknown | 91.5[77–99] | |||||||||||||||
| 262 | Phi-1.5Open weights | Microsoft | 91.4[70–101] | |||||||||||||||
| 263 | Baichuan1-7BOpen weights | Baichuan | 90.4[79–99] | |||||||||||||||
| 264 | RedPajama-INCITE-7B-Base | Unknown | 90.3[75–98] | |||||||||||||||
| 265 | Dolly 2.0-12bOpen weights | Databricks | 89.6[75–97] | |||||||||||||||
| 266 | DeepSeek Coder 6.7BOpen weights | DeepSeek,Peking University | 89.5[75–97] | |||||||||||||||
| 267 | StarCoder 2 3BOpen weights | Hugging Face,ServiceNow,Nvidia,BigCode | 88.7[73–98] | |||||||||||||||
| 268 | Qwen2.5-Coder-0.5B | Unknown | 88.0[62–100] | |||||||||||||||
| 269 | Cerebras-GPT-13BOpen weights | Cerebras Systems | 83.1[68–91] | |||||||||||||||
| 270 | DeepSeek Coder 1.3BOpen weights | DeepSeek,Peking University | 63.1[47–78] | |||||||||||||||
| 271 | stablelm-tuned-alpha-7b | Unknown | 55.5[31–72] |
Source: Epoch AI Benchmarking Hub · CC BY 4.0 · Last updated · All sources
Citation: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks. Download the data (JSON, CC BY 4.0).