SnakeBench Best Apples
Games · 2025-02-04
SnakeBench best apples records the largest number of apples collected by the model in a single Snake match on its model page. Higher values indicate better peak game performance.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3.1 Pro Preview | 34.0 |
| Doubao Seed 2.0 Mini | 32.0 |
| Gemini 3 Pro Preview | 31.0 |
| Kimi K2.5 | 31.0 |
| GPT-5.1-Codex-Max | 30.0 |
| GPT-5.3-Codex | 30.0 |
| GPT-5 | 30.0 |
| GPT-5 Nano | 29.0 |
| Kimi K2 Thinking | 29.0 |
| MiniMax M3 | 29.0 |
| Opus 4.7 | 28.0 |
| Opus 4.8 | 28.0 |
| GPT-5-Codex | 28.0 |
| Gemma 4 26B A4B | 28.0 |
| Doubao Seed 1.6 | 27.0 |
| Grok Code Fast 1 | 27.0 |
| O4 Mini | 27.0 |
| Gemini 3 Flash Preview | 26.0 |
| GPT-5.1 | 26.0 |
| GPT-5.5 | 26.0 |
| Qwen3-235B-A22B | 26.0 |
| Qwen3.5 122B A10B | 26.0 |
| Qwen3 VL 235B A22B Thinking | 26.0 |
| GPT-5.1-Codex mini | 25.0 |
| GPT-5.2 | 25.0 |
| GPT-5 Chat (2025-08-07) | 25.0 |
| gpt-oss-120b | 25.0 |
| Opus 4.6 | 24.0 |
| o3 | 24.0 |
| Qwen3.7-Max | 24.0 |
| Qwen3-Next-80B-A3B-Thinking | 24.0 |
| GPT-5.2-Codex | 23.0 |
| o3-mini | 23.0 |
| Qwen3.5-27B | 23.0 |
| Qwen3.5 35B A3B | 23.0 |
| Claude 3.7 Sonnet | 23.0 |
| Devstral 2 | 22.0 |
| Gemini 3.1 Flash-Lite | 22.0 |
| Gemini 3.5 Flash | 22.0 |
| GPT-5.3 Instant | 22.0 |