RuneBench Skill - Fletching
Games · 2026-02-26
RuneBench 30-minute Fletching skill task. Agents play RuneScape through the SDK and are scored by the best real-game Fletching XP per minute achieved in any 15-second window.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.5 | 1041 |
| Grok 4.5 | 937 |
| Opus 4.8 | 758 |
| Claude Fable 5 | 637 |
| Gemini 3.5 Flash | 498 |
| GPT-5.6 Luna | 450 |
| GPT-5.4 | 374 |
| Gemini 3 Flash Preview | 332 |
| Opus 4.7 | 225 |
| Qwen3.7-Max | 225 |
| GLM-5.2 | 221 |
| GPT-5.6 Sol | 175 |
| Opus 4.6 | 120 |
| Gemini 3.1 Pro Preview | 120 |
| Sonnet 5 | 100 |
| Opus 4.5 | 94.0 |
| Kimi K2.7 Code | 90.0 |
| DeepSeek-V4-Pro | 79.0 |
| Gemini 3 Pro Preview | 67.0 |
| GPT-5.4 Mini | 64.0 |
| GPT-5.4 Nano | 45.0 |
| Sonnet 4.6 | 41.0 |
| Kimi K2.6 | 38.0 |
| Grok 4.3 | 26.0 |
| Qwen3 Max (rolling alias) | 7.0 |
| Sonnet 4.5 | 4.0 |
| GPT-5.3-Codex | 4.0 |
| Haiku 4.5 | 0.0 |
| GLM-5 | 0.0 |
| Kimi K2.5 | 0.0 |
| Qwen3.5 35B A3B | 0.0 |