Atlas

Benchmarks

← All benchmarks

Agents' Last Exam v1.1 Score

Agents

Average partial-credit rubric score over all 122 Agents' Last Exam v1.1 release tasks, reported separately from the stricter perfect-score pass rate; a failed run scores zero.

Top models (higher is better)

ModelScore
GPT-6 Astra70.3
Claude Opus 5.567.5
Muse Wonder66.7
Gemini 4 Argon63.8
Kimi K362.5
MiniMax M3.1 Flash58.2
GLM-5.353.2
GPT-5.6 Sol48.4
GPT-5.6 Luna26.5
Loading Atlas data…