Vending bench 2 — Epoch AI
Vending bench 2 — Epoch AI is run by Epoch AI. It has ranked 60 models, of which the catalogue holds 54, scoring from -3118.4 to 11181.873.
measured by Epoch AI · the board itself
| Place | Model | Metric | Score |
|---|---|---|---|
| 1stof 60 | Claude Opus 5 | Score | 11181.873 |
| 2ndof 60 | Claude Opus 4.7 | Score | 10936.763 |
| 3rdof 60 | GPT-5.6 Sol | Score | 9619.368 |
| 4thof 60 | Grok 4.6 | Score | 9047.033 |
| 5thof 60 | GLM 5.2 | Score | 8313.778 |
| 6thof 60 | GLM-5.3 | Score | 8163.61 |
| 7thof 60 | Claude Opus 4.6 | Score | 8017.594 |
| 8thof 60 | GPT-5.5 | Score | 7523.844 |
| 9thof 60 | GPT-5.6 Terra | Score | 7343.206 |
| 10thof 60 | Claude Sonnet 4.6 | Score | 7204.14 |
| 11thof 60 | Muse Spark 1.1 | Score | 6520.475 |
| 12thof 60 | Claude Sonnet 5 | Score | 6377.702 |
| 13thof 60 | Kimi K2.6 | Score | 6204.57 |
| 14thof 60 | GPT-5.4 | Score | 6144.176 |
| 15thof 60 | GPT-5.3 Codex | Score | 5940.124 |
| 16thof 60 | Claude Opus 4.8 | Score | 5787.428 |
| 17thof 60 | Claude Fable 5 | Score (high) | 5680.258 |
| 18thof 60 | GLM 5.1 | Score | 5634.406 |
| 19thof 60 | gemini-3-pro-preview | Score | 5478.158 |
| 20thof 60 | Claude Fable 5.1 | Score | 5421.557 |
| 21stof 60 | Gemini 3.5 Flash | Score | 5396.418 |
| 22ndof 60 | Kimi K3 | Score | 5165.035 |
| 23rdof 60 | Qwen3.6 Plus | Score | 5114.872 |
| 24thof 60 | Kimi K2.7 Code | Score | 5082.936 |
| 26thof 60 | Claude Opus 4.5 | Score | 4967.057 |
| 28thof 60 | Grok 4.20 | Score | 4662.847 |
| 30thof 60 | GLM-5 | Score | 4432.122 |
| 32ndof 60 | Qwen3.6 Max Preview | Score (max) | 4254.187 |
| 33rdof 60 | GPT-5.6 Luna | Score | 4094.712 |
| 34thof 60 | Grok 4.5 | Score | 3887.426 |
| 35thof 60 | Claude Sonnet 4.5 | Score | 3838.74 |
| 36thof 60 | Gemini 3.1 Pro Preview Custom Tools | Score | 3774.254 |
| 37thof 60 | Gemini 3 Flash Preview | Score | 3634.716 |
| 38thof 60 | GPT-5.2 | Score | 3591.326 |
| 39thof 60 | DeepSeek V4 Pro | Score | 3284.52 |
| 41stof 60 | GLM-4.7 | Score | 2376.816 |
| 42ndof 60 | MiniMax M3 | Score | 2157.772 |
| 43rdof 60 | GPT-5.1 | Score | 1473.432 |
| 44thof 60 | Kimi K2.5 | Score | 1198.458 |
| 45thof 60 | Grok 4.1 Fast Reasoning | Score | 1106.628 |
| 46thof 60 | DeepSeek V3.2 Exp | Score | 1034.004 |
| 47thof 60 | Gemini 3.1 Pro Preview | Score | 911.214 |
| 48thof 60 | Gemini 2.5 Pro | Score | 573.642 |
| 49thof 60 | Gemini 2.5 Flash | Score | 548.836 |
| 50thof 60 | Qwen3.5-Flash | Score | 462.685 |
| 51stof 60 | Claude Haiku 4.5 | Score | 458.886 |
| 52ndof 60 | Qwen3.5-27B | Score | 201.976 |
| 53rdof 60 | MiniMax-M2 | Score | 160.596 |
| 55thof 60 | Qwen 3.5 Plus | Score | 54.4 |
| 56thof 60 | Grok 4.3 | Score | 35.262 |
| 57thof 60 | Qwen3 235B A22B Thinking 2507 | Score | -1134 |
| 58thof 60 | GPT OSS 120B | Score | -2153.2 |
| 59thof 60 | MiniMax M2.5 | Score | -2316 |
| 60thof 60 | GPT-5 mini | Score | -3118.4 |
Read from the board on 2026-09-16