AI RESEARCH

Medmarks: A Comprehensive Open-Source LLM Benchmark Suite for Medical Tasks

arXiv CS.CL

ArXi:2605.01417v1 Announce Type: new Evaluating large language models (LLMs) for medical applications remains challenging due to benchmark saturation, limited data accessibility, and insufficient coverage of relevant tasks. Existing suites have either saturated, heavily depend on restricted datasets, or lack comprehensive model coverage. We