Home Model series Claude Opus Claude Opus 5 Claude Opus 5 · Legal Agent Benchmark · Opus 5 官方发布表所列 held-out set · Scores and configurations Claude Opus 5 · Legal Agent Benchmark · Opus 5 官方发布表所列 held-out set · Scores and configurations Claude Opus 5 record for Legal Agent Benchmark · Opus 5 官方发布表所列 held-out set. 2 score records for this release, including 2 exact values (may include historical records). Includes 2 source locations.
Scores and evaluation conditions Claude Opus 5 · Legal Agent Benchmark scores and configurations Metric and score Configuration and evaluation conditions Sources All-pass rate 11.7% Reported 11.7% · Unit % Published Unconfirmed
Legal Agent Benchmark · Harvey held-out set Settings External Harvey evaluation on its held-out set; effort, API, toolset, and run count not specified in the system card. Unknowns Do not combine with Anthropic internal full-set evaluation. Tools Tool access not recorded Reasoning budget not recorded Framework not recorded Protocol Legal Agent Benchmark; Harvey held-out set; all-pass rate. Conditions External Harvey evaluation; effort/API/tool details not stated; separate from the Anthropic full-set result. View this score in rankings Mean criterion-pass rate 94.1% Reported 94.1% · Unit % Published Unconfirmed
Legal Agent Benchmark · Harvey held-out set Settings External Harvey evaluation on its held-out set; effort, API, toolset, and run count not specified in the system card. Unknowns Do not combine with Anthropic internal full-set evaluation. Tools Tool access not recorded Reasoning budget not recorded Framework not recorded Protocol Legal Agent Benchmark; Harvey held-out set; mean criterion-pass rate. Conditions External Harvey evaluation; effort/API/tool details not stated; separate from the Anthropic full-set result. View this score in rankings
Benchmark versions and metrics Aliases Legal Agent Benchmark (held-out)
All-pass rate · % · Higher is better Mean criterion-pass rate · % · Higher is better