Benchmark RadarAI BENCHMARK PROFILE
HealthBench-Psych
Evaluates LLMs on 610 mental-health conversations from HealthBench using a three-judge panel and physician rubrics, with a harder 119-conversation subset.
- Released
- 2026-08-25
- Readiness
- Runnable
- Primary field
- Health & Life Sciences
Why it matters
Provides a clinically validated mental-health evaluation subset with released data, code, and judge panel, enabling reproducible specialty-specific model assessment.
Motivation
General-purpose health benchmarks increasingly anchor claims about LLM medical performance, but they are not always resolved by clinical specialty, making domain-specific performance hard to isolate.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.