Ale bench — Epoch AI
Ale bench — Epoch AI is run by Epoch AI. It has ranked 112 models, of which the catalogue holds 98, scoring from 137.78 to 2176.88.
measured by Epoch AI · the board itself
| Place | Model | Metric | Score |
|---|---|---|---|
| 1stof 112 | GPT-5.6 Sol | Performance (max) | 2176.88 |
| 2ndof 112 | Claude Opus 5 | Performance (high) | 2164.57 |
| 3rdof 112 | Claude Fable 5.1 | Performance (high) | 2143.2 |
| 4thof 112 | Claude Fable 5 | Performance (high) | 2041.3 |
| 5thof 112 | GPT-5.6 Terra | Performance (max) | 1951.42 |
| 6thof 112 | GPT-5.5 | Performance (xhigh) | 1942.97 |
| 7thof 112 | GPT-5.6 Luna | Performance (max) | 1667.4 |
| 8thof 112 | GPT-5.3 Codex | Performance (xhigh) | 1655.22 |
| 9thof 112 | GPT-5.4 | Performance (high) | 1607 |
| 11thof 112 | Claude Opus 4.8 | Performance (high) | 1563.83 |
| 12thof 112 | Kimi K3 | Performance (max) | 1524.47 |
| 14thof 112 | Grok 4.6 | Performance (xhigh) | 1508.08 |
| 15thof 112 | Claude Sonnet 5 | Performance (high) | 1463.12 |
| 17thof 112 | DeepSeek V4 Pro 0813 | Performance (max) | 1403.15 |
| 18thof 112 | Gemini 3 Flash Preview | Performance | 1367.2 |
| 19thof 112 | Claude Sonnet 4.6 | Performance (medium) | 1327.3 |
| 20thof 112 | Claude Opus 4.7 | Performance | 1323.05 |
| 21stof 112 | GLM-5.3 | Performance (high) | 1317.4 |
| 22ndof 112 | Grok 4.5 | Performance (high) | 1309.28 |
| 23rdof 112 | DeepSeek V4 Flash (0731) | Performance (max) | 1306.08 |
| 24thof 112 | GPT-5.2 Codex | Performance | 1299.9 |
| 25thof 112 | GPT-5.2 | Performance (high) | 1293.55 |
| 27thof 112 | GPT-5.1 Codex | Performance | 1244.92 |
| 28thof 112 | GPT-5.1 Codex Max | Performance (max) | 1208.83 |
| 29thof 112 | GPT-5.1 | Performance (high) | 1192.15 |
| 30thof 112 | Qwen3.7 Max | Performance (max) | 1189.4 |
| 31stof 112 | GPT-5.4 mini | Performance (high) | 1188.58 |
| 32ndof 112 | gemini-3-pro-preview | Performance | 1176.75 |
| 33rdof 112 | GPT-5 | Performance (high) | 1162.45 |
| 34thof 112 | Gemini 3.1 Pro Preview | Performance | 1160.6 |
| 35thof 112 | Grok 4.20 | Performance | 1150.28 |
| 37thof 112 | Kimi K2.6 | Performance | 1092.67 |
| 39thof 112 | GLM 5.2 | Performance (high) | 1046.95 |
| 40thof 112 | Claude Opus 4.5 | Performance | 1025.38 |
| 42ndof 112 | DeepSeek V4 Pro | Performance (high) | 1006.08 |
| 43rdof 112 | GPT-5.4 nano | Performance (high) | 1004.52 |
| 44thof 112 | Claude Opus 4.6 | Performance | 996.5 |
| 45thof 112 | Inkling | Performance | 946 |
| 46thof 112 | Grok 4.3 | Performance | 944.17 |
| 47thof 112 | o3 | Performance (high) | 933.55 |
| 48thof 112 | Gemma 4 26B A4B | Performance | 927.17 |
| 49thof 112 | Gemma 4 31B | Performance | 925.5 |
| 50thof 112 | Gemini 3.5 Flash | Performance (high) | 911.02 |
| 51stof 112 | Gemini 3.7 Flash | Performance (high) | 904.3 |
| 52ndof 112 | MiMo-V2.5-Pro | Performance | 899.8 |
| 53rdof 112 | GLM 5.1 | Performance | 887.1 |
| 54thof 112 | Kimi K2.7 Code | Performance | 886.23 |
| 55thof 112 | o4-mini | Performance (high) | 826.17 |
| 56thof 112 | Kimi K2.5 | Performance | 821.65 |
| 58thof 112 | DeepSeek-R1 (0528) | Performance | 804.12 |
| 59thof 112 | GPT-5 mini | Performance (high) | 799.77 |
| 60thof 112 | Gemini 3.1 Flash-Lite | Performance | 797.73 |
| 61stof 112 | Claude Sonnet 4.5 | Performance | 796.15 |
| 62ndof 112 | Mercury 2 | Performance | 785.58 |
| 63rdof 112 | Gemini 2.5 Pro | Performance | 785.52 |
| 64thof 112 | MiMo-V2-Pro | Performance | 785.17 |
| 65thof 112 | GLM-5 | Performance | 765.62 |
| 66thof 112 | Gemini 3.5 Flash-Lite | Performance (high) | 765.27 |
| 68thof 112 | DeepSeek V3.1 Terminus | Performance | 745.17 |
| 69thof 112 | MiMo-V2-Flash | Performance | 737.95 |
| 70thof 112 | GPT-5 nano | Performance (high) | 718.67 |
| 71stof 112 | Gemini 3.6 Flash | Performance (high) | 715.52 |
| 72ndof 112 | Step 3.7 Flash | Performance | 694.12 |
| 73rdof 112 | DeepSeek V4 Flash | Performance (high) | 678.2 |
| 74thof 112 | Claude Opus 4.1 | Performance | 674.77 |
| 75thof 112 | Qwen3.6 Plus | Performance | 670.15 |
| 76thof 112 | Gemini 2.5 Flash | Performance | 661.88 |
| 77thof 112 | Claude Sonnet 4 | Performance | 655.35 |
| 78thof 112 | Claude Haiku 4.5 | Performance | 653.48 |
| 79thof 112 | MiniMax M3 | Performance | 640.02 |
| 80thof 112 | MiniMax M2.1 | Performance | 623.83 |
| 81stof 112 | Qwen 3.5 Plus | Performance | 621.92 |
| 82ndof 112 | MiniMax M2.5 | Performance | 618.17 |
| 83rdof 112 | MiniMax M2.7 | Performance | 599.25 |
| 84thof 112 | Kimi K2 Thinking | Performance | 597.5 |
| 85thof 112 | grok-code-fast-1 | Performance | 587.73 |
| 86thof 112 | GPT OSS 120B | Performance | 575.62 |
| 87thof 112 | GPT OSS 20B | Performance | 566.05 |
| 88thof 112 | GPT-4.1 fine-tuned | Performance | 558.1 |
| 90thof 112 | MiMo-V2.5 | Performance | 513.95 |
| 91stof 112 | Mistral Small 4 | Performance | 497.62 |
| 92ndof 112 | Qwen3-Coder-480B-A35B-Instruct | Performance | 461.45 |
| 93rdof 112 | Qwen3-Coder-Plus | Performance | 456.5 |
| 94thof 112 | Ring-2.6-1T | Performance | 432.57 |
| 95thof 112 | GLM-4.7 | Performance | 399.48 |
| 96thof 112 | Grok 4.1 Fast Reasoning | Performance | 394.93 |
| 98thof 112 | Qwen3.5-27B | Performance | 349.45 |
| 99thof 112 | GLM-4.5 | Performance | 344.82 |
| 100thof 112 | GLM-4.6 | Performance | 340.82 |
| 101stof 112 | Qwen3.6-Flash | Performance | 326.4 |
| 104thof 112 | GLM 5.3 Flash | Performance (high) | 303.55 |
| 105thof 112 | Kimi K2 0905 | Performance | 267.12 |
| 106thof 112 | Mistral Large 3 | Performance | 264.7 |
| 107thof 112 | Nova Lite | Performance | 236.25 |
| 108thof 112 | Qwen3.5-Flash | Performance | 221.8 |
| 109thof 112 | Nemotron 3 Super | Performance | 213.9 |
| 111thof 112 | Llama 4 Maverick 17B | Performance | 172.97 |
| 112thof 112 | Codestral 2508 | Performance | 137.78 |
Read from the board on 2026-09-16