RuneBench Skill - Thieving
Games · 2026-02-26
RuneBench 30-minute Thieving skill task. Agents play RuneScape through the SDK and are scored by the best real-game Thieving XP per minute achieved in any 15-second window.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6 Astra | 5268 |
| Claude Opus 5.5 | 5142 |
| GPT-6.1 Sol | 4510 |
| Claude Opus 5 | 4341 |
| Grok 4.6 | 3161 |
| GPT-5.6 Terra | 2201 |
| Claude Fable 5.1 | 2023 |
| Gemini 3.7 Flash | 1990 |
| Gemini 4 Argon | 1913 |
| GPT-5.5 | 1640 |
| GPT-5.6 Sol | 1503 |
| GPT-6 Sol | 1503 |
| Opus 4.7 | 1261 |
| Grok 4.7 | 1012 |
| Claude Sonnet 5.5 | 983 |
| Gemini 3.5 Flash | 889 |
| GPT-6 Luna | 845 |
| Claude Fable 5 | 835 |
| Grok 4.5 | 796 |
| DeepSeek V4.1 Flash | 780 |
| Gemini 3.8 Flash | 702 |
| Muse Spark 1.2 | 548 |
| GLM-5.3 | 495 |
| Gemini 3.6 Flash | 445 |
| Opus 4.8 | 442 |
| GPT-5.6 Luna | 364 |
| GPT-5.3-Codex | 280 |
| Claude Haiku 5.5 | 276 |
| GLM-5.3 Flash | 260 |
| Muse Wonder | 236 |
| Muse Spark 1.1 | 234 |
| Opus 4.5 | 232 |
| Kimi K3 | 221 |
| GLM-5 | 208 |
| Sonnet 5 | 192 |
| Qwen3.7-Max (source unspecified) | 192 |
| DeepSeek-V4-Flash-0731 | 188 |
| GLM-5.2 | 187 |
| Grok 4.3 | 172 |
| Muse Spark 1.3 | 168 |