About

Clinical Evals explains how healthcare AI evaluation works. It carries no rankings and takes no position on which model to buy; its job is the machinery, so that when you read a score anywhere, you know what produced it.

Every number on this site comes from a benchmark's published paper or its named source, and pages link to primary material wherever it exists. Where a claim cannot be sourced, it is not made.

The benchmark facts on these pages, and the scores on the sister sites listed below, are read from one shared results database that the whole family maintains. Every score in it is stored together with the document it was taken from, and clinicalbenchmarks.ai publishes that source for each number, so you can check any figure you meet against its origin.

This site is run by the same team as clinicalbenchmarks.ai, the results index, and the dedicated leaderboards at healthbenchprofessional.com, healthbenchhard.ai, and healthoptimizationbench.com. Guides here point at those boards for current standings; nothing on this site duplicates them. Questions and corrections: contact.