RuneBench Skill - Crafting
Games · 2026-02-26
RuneBench 30-minute Crafting skill task. Agents play RuneScape through the SDK and are scored by the best real-game Crafting XP per minute achieved in any 15-second window.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 1225 |
| Claude Opus 5 | 814 |
| GPT-6.1 Sol | 805 |
| Grok 4.6 | 787 |
| Claude Fable 5 | 770 |
| GPT-6 Astra | 758 |
| Gemini 4 Argon | 745 |
| Claude Fable 5.1 | 718 |
| GPT-6 Sol | 703 |
| GPT-5.6 Sol | 438 |
| GPT-5.6 Terra | 437 |
| Grok 4.7 | 403 |
| Gemini 3.8 Flash | 245 |
| Gemini 3.7 Flash | 210 |
| Grok 4.5 | 152 |
| GPT-5.5 | 151 |
| GPT-6 Luna | 131 |
| Gemini 3.5 Flash | 130 |
| DeepSeek V4.1 Flash | 125 |
| Gemini 3 Flash Preview | 117 |
| GPT-5.6 Luna | 113 |
| Gemini 3.6 Flash | 110 |
| Opus 4.8 | 106 |
| GLM-5.2 | 103 |
| Muse Spark 1.3 | 97.0 |
| GPT-5.4 | 84.0 |
| GPT-5.3-Codex | 76.0 |
| Muse Spark 1.2 | 76.0 |
| DeepSeek-V4-Flash-0731 | 75.0 |
| GLM-5.3 Flash | 69.0 |
| Opus 4.6 | 62.0 |
| Muse Spark 1.1 | 55.0 |
| Claude Sonnet 5.5 | 35.0 |
| DeepSeek-V4-Pro | 28.0 |
| Muse Wonder | 27.0 |
| Sonnet 5 | 26.0 |
| Claude Haiku 5.5 | 16.0 |
| Kimi K2.7 Code | 7.0 |
| Opus 4.7 | 1.0 |
| Sonnet 4.6 | 1.0 |