Updates

The index's history, newest first: benchmarks added or retired, results refreshed, corrections made. Anchors are permanent.

• 2026-09-28

  Health Evals now focuses on MedPIC-Bench, a public medication-safety benchmark. Added all 28 published model results, guideline-following and counterfactual comparisons, linked-pair scores, dataset coverage views, and dedicated JSON/CSV downloads. Scores come from the 4 August 2026 paper; the review date does not represent a new evaluation run. The broader healthcare AI index remains available under All benchmarks. Added an original visual identity with vector artwork.

• 2026-09-28

  Health Evals launched with 17 healthcare AI benchmark boards, adapted from the Clinical Benchmarks index. Sources reviewed for newer results, with per-board review status, per-result citations, and original measurement dates retained. Added benchmark search, category filters, model pages, and JSON/CSV downloads. The source review notes on each board identify additions, corrections, and verification limits.

The broader collection is on the benchmark index, and the current snapshot is downloadable from the data page.