Benchmark RadarAI BENCHMARK PROFILE
CARE-Bench
CARE-Bench is a source-grounded benchmark for patient-facing triage evaluation with 500 cases and 1,059 patient-disclosure prefixes. It evaluates LLMs on a four-label current-action task across 269 held-out rounds.
- Released
- 2026-08-04
- Readiness
- Paper only
- Primary field
- Health & Life Sciences
Why it matters
Provides a standardized evaluation for patient-facing triage, enabling comparison of models on action timing and clarification steps, critical for deployment safety.
Motivation
Patient-facing medical LLMs and agents increasingly answer symptom questions before clinician contact, where the key safety question is what action the user should take next.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.