RuneBench Skill - Strength
Games · 2026-02-26
RuneBench 30-minute Strength skill task. Agents play RuneScape through the SDK and are scored by the best real-game Strength XP per minute achieved in any 15-second window.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.6 Sol | 356 |
| Grok 4.5 | 334 |
| Claude Fable 5 | 286 |
| Opus 4.8 | 210 |
| Sonnet 4.5 | 204 |
| Kimi K2.7 Code | 202 |
| Gemini 3 Flash Preview | 198 |
| Kimi K2.6 | 194 |
| GPT-5.5 | 184 |
| Gemini 3.5 Flash | 180 |
| GPT-5.6 Luna | 180 |
| Qwen3.7-Max | 178 |
| Gemini 3.1 Pro Preview | 176 |
| DeepSeek-V4-Pro | 166 |
| Opus 4.6 | 162 |
| GLM-5.2 | 138 |
| Grok 4.3 | 136 |
| Sonnet 4.6 | 132 |
| Sonnet 5 | 132 |
| Opus 4.5 | 112 |
| GPT-5.4 Mini | 106 |
| Opus 4.7 | 96.0 |
| Gemini 3 Pro Preview | 96.0 |
| GPT-5.3-Codex | 80.0 |
| GPT-5.4 | 64.0 |
| Kimi K2.5 | 56.0 |
| GPT-5.4 Nano | 42.0 |
| Haiku 4.5 | 0.0 |
| GLM-5 | 0.0 |
| Qwen3.5 35B A3B | 0.0 |
| Qwen3 Max (rolling alias) | 0.0 |