Atlas

Benchmarks

← All benchmarks

RuneBench Skill - Magic

Games · 2026-02-26

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

Top models (higher is better)

ModelScore
Gemini 4 Argon2061
GPT-6 Astra1820
GPT-6.1 Sol1788
GPT-6 Sol675
Claude Opus 5.5604
Grok 4.6421
Claude Sonnet 5.5320
Grok 4.5237
GLM-5.2203
GPT-5.6 Sol179
Claude Opus 5177
GPT-5.6 Terra156
DeepSeek V4.1 Flash147
GPT-6 Luna145
Opus 4.8130
Opus 4.6100
Claude Fable 5.199.0
Muse Wonder96.0
GPT-5.3-Codex90.0
Grok 4.790.0
Claude Haiku 5.585.0
Muse Spark 1.181.0
GPT-5.580.0
Sonnet 579.0
Muse Spark 1.377.0
Gemini 3.8 Flash76.0
Gemini 3.5 Flash74.0
Claude Fable 572.0
Gemini 3.6 Flash72.0
Opus 4.762.0
Gemini 3 Flash Preview53.0
GLM-548.0
Sonnet 4.546.0
GPT-5.446.0
DeepSeek-V4-Pro43.0
GPT-5.4 Mini39.0
Gemini 3.7 Flash33.0
GPT-5.6 Luna33.0
Mistral Large 4 Preview32.0
Gemini 3.1 Pro Preview31.0
Loading Atlas data…