Home Model series Claude Opus Claude Opus 4.6 Claude Opus 4.6 · Humanity's Last Exam · 2,500-question benchmark · Scores and configurations Claude Opus 4.6 · Humanity's Last Exam · 2,500-question benchmark · Scores and configurations Claude Opus 4.6 record for Humanity's Last Exam · 2,500-question benchmark. 2 score records for this release, including 2 exact values (may include historical records). Includes 4 source locations.
Scores and evaluation conditions Claude Opus 4.6 · Humanity's Last Exam scores and configurations Metric and score Configuration and evaluation conditions Sources Accuracy 40% Reported 40.0% · Unit % Published Unconfirmed
HLE 无工具 Settings Humanity’s Last Exam;无工具;卡片图注明确 adaptive thinking,但未明确 effort。 Unknowns effort 未说明。 Tools Tools forbidden Reasoning budget not recorded Framework not recorded Protocol Humanity's Last Exam;without tools。 Conditions 2,500-question Humanity’s Last Exam;reasoning only, no tools;adaptive thinking enabled;图注未明确 effort,保持 reported。 View this score in rankings Accuracy 53% Reported 53.0% · Unit % Published 2026-02-23
raw/opus-4-6-announcement.txt line 138;官方修订说明。
HLE 有工具 Settings Max effort、adaptive thinking;web search/fetch、programmatic tool calling、code execution;50K context compaction,最多 3M total tokens。 Unknowns 每个配置的试验数未在系统卡段落中给出。 Tools Tools allowed Reasoning budget max Framework not recorded Protocol Humanity's Last Exam;with tools。 Conditions 2,500-question HLE;web search/fetch、PTC、code execution、50K context compaction up to 3M tokens;adaptive thinking、max effort;Sonnet 4.5 grader;污染候选经 transcript review 与人工复核。System card changelog dates the score correction Feb. 17; announcement correction evidence dates it Feb. 23. View this score in rankings
Benchmark versions and metrics Aliases HLE、Humanity’s Last Exam、Humanity's Last Exam
Accuracy · % · Higher is better