SnakeBench Rating
Games · 2025-02-04
SnakeBench rating is the leaderboard rating assigned to a model from its Snake matches. Higher ratings indicate stronger model performance in the Snake game setting.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3 Pro Preview | 38.0 |
| GPT-5.5 | 37.4 |
| Gemini 3.1 Pro Preview | 37.0 |
| GPT-5.1-Codex mini | 36.5 |
| o3 | 36.4 |
| GPT-5.1-Codex-Max | 36.4 |
| Opus 4.8 | 35.8 |
| GPT-5 | 35.0 |
| Qwen3.7-Max | 34.1 |
| O4 Mini | 33.9 |
| GPT-5-Codex | 33.3 |
| Grok 4 Fast | 33.2 |
| GPT-5.2 | 33.2 |
| Gemini 2.0 Flash 001 | 33.2 |
| Gemini 3.5 Flash | 33.2 |
| Qwen3.5 122B A10B | 33.0 |
| GPT-5.1 | 33.0 |
| Sonnet 4.5 | 33.0 |
| Opus 4.6 | 32.9 |
| GPT-4.1 | 32.9 |
| GPT-5.3-Codex | 32.7 |
| Sonnet 4 | 32.7 |
| Qwen3 VL 30B A3B Thinking | 32.5 |
| GPT-5.2-Codex | 32.4 |
| Qwen3.5-Flash | 32.4 |
| GPT-5.1-Codex | 32.1 |
| Gemini 3 Flash Preview | 31.9 |
| Mercury | 31.9 |
| Qwen3.6-Max-Preview | 31.6 |
| Gemini 2.5 Flash | 31.3 |
| DeepSeek-V3.1-Terminus | 31.2 |
| Qwen3.5-27B | 30.9 |
| Qwen3 Max (rolling alias) | 30.8 |
| Gemini 2.5 Flash Preview (09-2025) | 30.8 |
| Gemini 3.1 Flash-Lite Preview | 30.7 |
| GLM-5 | 30.5 |
| DeepSeek-R1-Distill-Llama-70B | 30.3 |
| Opus 4.5 | 30.1 |
| Opus 4.7 | 30.1 |
| Kimi K2 Instruct 0905 | 30.1 |