Atlas

Benchmarks

← All benchmarks

CUA-bench — Hidden Sandbox

Agents · 2026-09-18

Ordered-milestone progress on the held-out sandbox game within three hours.

Top models (higher is better)

ModelScore
Claude Opus 5.520.0
GPT-6 Astra15.0
Claude Fable 5.10.0
Claude Opus 50.0
Gemini 3.8 Flash0.0
Gemini 4 Argon0.0
GPT-5.6 Sol0.0
Muse Spark 1.30.0
Muse Wonder0.0
Loading Atlas data…