Benchmark RadarDLawBench
DLawBench evaluates multi-turn legal consultation in Chinese and U.S. law across four client personas. It scores LLMs on information gathering, legal reasoning, and claim support using 461 cases, 5,532 paired fact entries, inquiry and issue rubrics, and a public leaderboard.
- Released
- 2026-06-11
- Readiness
- Runnable
- Primary field
- General AI
Why it matters
Existing legal benchmarks assume complete fact patterns, overlooking the interactive elicitation needed in real consultations. DLawBench provides a diagnostic evaluation of LLM capability in legal consultation, revealing performance gaps and failure modes that inform development of models for legal assistance.
Motivation
Lawyer-client consultation is a critical starting point for legal services.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.