Atlas

Benchmarks

← All benchmarks

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)

ModelScore
GPT-6 Astra5268
Claude Opus 5.55142
GPT-6.1 Sol4510
Claude Opus 54341
Grok 4.63161
GPT-5.6 Terra2201
Claude Fable 5.12023
Gemini 3.7 Flash1990
Gemini 4 Argon1913
GPT-5.51640
GPT-5.6 Sol1503
GPT-6 Sol1503
Opus 4.71261
Grok 4.71012
Claude Sonnet 5.5983
Gemini 3.5 Flash889
GPT-6 Luna845
Claude Fable 5835
Grok 4.5796
DeepSeek V4.1 Flash780
Gemini 3.8 Flash702
Muse Spark 1.2548
GLM-5.3495
Gemini 3.6 Flash445
Opus 4.8442
GPT-5.6 Luna364
GPT-5.3-Codex280
Claude Haiku 5.5276
GLM-5.3 Flash260
Muse Wonder236
Muse Spark 1.1234
Opus 4.5232
Kimi K3221
GLM-5208
Sonnet 5192
Qwen3.7-Max (source unspecified)192
DeepSeek-V4-Flash-0731188
GLM-5.2187
Grok 4.3172
Muse Spark 1.3168
Loading Atlas data…