Atlas

Benchmarks

← All benchmarks

RuneBench Skill - Cooking

Games · 2026-02-26

RuneBench 30-minute Cooking skill task. Agents play RuneScape through the SDK and are scored by the best real-game Cooking XP per minute achieved in any 15-second window.

Top models (higher is better)

ModelScore
GPT-5.6 Sol1470
GPT-5.51080
Grok 4.5905
Claude Fable 5785
Gemini 3 Flash Preview500
GPT-5.6 Luna375
Qwen3.7-Max375
Gemini 3.5 Flash360
Gemini 3.1 Pro Preview345
Opus 4.8270
GPT-5.3-Codex240
Sonnet 5210
GPT-5.4210
Opus 4.7195
GLM-5.2180
Kimi K2.7 Code135
GPT-5.4 Mini90.0
Sonnet 4.575.0
Opus 4.560.0
Grok 4.345.0
Opus 4.630.0
DeepSeek-V4-Pro30.0
Gemini 3 Pro Preview15.0
Kimi K2.615.0
Haiku 4.50.0
Sonnet 4.60.0
GLM-50.0
GPT-5.4 Nano0.0
Kimi K2.50.0
Qwen3.5 35B A3B0.0
Qwen3 Max (rolling alias)0.0
Loading Atlas data…