Benchmark profile
Legal Agent Benchmark all-pass rate — Anthropic harness (LAB all-pass (Anthropic harness))
Strict task success requiring every expert-written legal-work rubric criterion to pass.
Data verifiedBenchmark score on LAB all-pass (Anthropic harness) — July 24, 2026
BenchLM mirrors the published score view for LAB all-pass (Anthropic harness). Claude Opus 5 leads the public snapshot at 23.58%. BenchLM does not use these results to rank models overall.
Benchmark score table (1 model)
ScoreAbout LAB all-pass (Anthropic harness)
Year
2026
Tasks
1,235 legal-agent tasks
Format
All-criteria pass rate
Difficulty
Professional legal work
Section 8.13.3 reports five-run performance on 1,235 tasks using Anthropic's reduced-tool reimplementation, production safeguards, and fallback to Opus 4.8 on classifier refusal.
BenchLM freshness & provenance
Version
LAB all-pass (Anthropic harness) 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
FAQ
What does LAB all-pass (Anthropic harness) measure?
Strict task success requiring every expert-written legal-work rubric criterion to pass.
Which model scores highest on LAB all-pass (Anthropic harness)?
Claude Opus 5 by Anthropic currently leads with a score of 23.58% on LAB all-pass (Anthropic harness).
How many models are evaluated on LAB all-pass (Anthropic harness)?
1 AI models have been evaluated on LAB all-pass (Anthropic harness) on BenchLM.