SnakeBench Win Rate
Games · 2025-02-04
SnakeBench win rate is the percentage of completed Snake matches won by a model. It summarizes head-to-head game performance across the model page match set.
Top models (higher is better)
| Model | Score |
|---|---|
| DeepSeek-R1 | 100.0 |
| Gemini 2.5 Pro Preview 06-05 | 100.0 |
| GPT-5.1-Codex-Max | 96.9 |
| GPT-5.1-Codex | 92.3 |
| Qwen3.5 Plus (2026-02-15) | 92.3 |
| Gemini 3 Pro Preview | 90.9 |
| GPT-5.2-Codex | 90.9 |
| Gemini 3.1 Flash-Lite Preview | 90.0 |
| GPT-5.5 | 90.0 |
| Qwen3.5 122B A10B | 90.0 |
| o3-deep-research | 90.0 |
| DeepSeek-V4-Flash | 88.9 |
| Gemini 2.5 Flash Image Preview | 88.9 |
| GLM 4.7 Flash | 88.9 |
| MiniMax M2.5 | 88.9 |
| Qwen3.6-Max-Preview | 88.9 |
| Qwen3.7-Max | 88.9 |
| Gemini 2.5 Pro Preview 05-06 | 87.5 |
| Aion 2.0 | 86.7 |
| Gemini 3.1 Pro Preview | 86.2 |
| GPT-5.1-Codex mini | 85.7 |
| GPT-5 Mini | 84.6 |
| GPT-5.2 | 83.3 |
| o3 | 83.3 |
| Opus 4.7 | 81.8 |
| Mercury 2 | 81.8 |
| O4 Mini | 81.8 |
| Palmyra X5 | 81.8 |
| Qwen3 VL 30B A3B Thinking | 81.8 |
| Gemini 2.0 Flash 001 | 81.3 |
| Gemini 3.1 Flash-Lite | 80.0 |
| Gemini 3.5 Flash | 80.0 |
| o3-mini | 80.0 |
| Qwen3.5-27B | 80.0 |
| KAT-Coder-Pro V2 | 80.0 |
| GLM-5 | 78.6 |
| Grok 4 Fast | 78.3 |
| Doubao Seed 1.6 Flash 250715 | 77.8 |
| Qwen3.5-Flash | 76.5 |
| Gemini 2.5 Flash Preview (09-2025) | 75.0 |