RuneBench Skill - Woodcutting
Games · 2026-02-26
RuneBench 30-minute Woodcutting skill task. Agents play RuneScape through the SDK and are scored by the best real-game Woodcutting XP per minute achieved in any 15-second window.
Top models (higher is better)
| Model | Score |
|---|---|
| Kimi K2.7 Code | 1181 |
| Claude Fable 5 | 1148 |
| GPT-5.6 Sol | 1125 |
| GPT-5.5 | 1046 |
| Grok 4.5 | 875 |
| Gemini 3.5 Flash | 810 |
| GLM-5.2 | 754 |
| Opus 4.7 | 612 |
| Opus 4.6 | 574 |
| GPT-5.6 Luna | 574 |
| Gemini 3.1 Pro Preview | 540 |
| GPT-5.4 Mini | 472 |
| Opus 4.8 | 469 |
| GPT-5.4 | 375 |
| Grok 4.3 | 375 |
| Opus 4.5 | 350 |
| Gemini 3 Pro Preview | 338 |
| GPT-5.3-Codex | 338 |
| Sonnet 5 | 275 |
| Qwen3.7-Max | 275 |
| DeepSeek-V4-Pro | 263 |
| GLM-5 | 213 |
| Kimi K2.6 | 162 |
| Kimi K2.5 | 63.0 |
| Sonnet 4.5 | 50.0 |
| Sonnet 4.6 | 38.0 |
| Haiku 4.5 | 25.0 |
| Gemini 3 Flash Preview | 25.0 |
| GPT-5.4 Nano | 25.0 |
| Qwen3.5 35B A3B | 13.0 |
| Qwen3 Max (rolling alias) | 0.0 |