Artificial Analysis Launches Harvey Legal Agent Benchmark

ArtificialAnlys · x · 2026-07-08

Artificial Analysis introduced Harvey LAB-AA (Legal Agent Benchmark), their implementation of Harvey's new agentic legal benchmark. It evaluates language models on real-world legal work across 24 practice areas.

The benchmark tests models on 120 private legal tasks built by the Harvey team, covering M&A, capital markets, tax, litigation, and bankruptcy. Models must produce specified legal deliverables, with each task scored on a binary scale.

Related event: Artificial Analysis Releases Harvey Legal Agent Benchmark Results(8 posts)→

Original post →

More from Research

Research channel →