MedR Bench
Fresh
MedR-Bench is a benchmark for evaluating reasoning-enhanced LLMs in clinical settings, comprising 1,453 structured patient cases across 13 body systems and 10 specialties annotated with reasoning references derived from clinical case reports.
- Type
- RL Env
- Runtime
ORS- License
- unknown
- Size
- 1453 tasks
- Published
- Feb 2026
- Canonical
- openreward.ai/Pengcheng/MedR-Bench
Cite
Notes
Only stored in your browser.