How six open-source FHIR R4 servers stack up against the spec — HAPI FHIR, Microsoft FHIR Server, Medplum, Aidbox, Blaze, and Spark. Independent. Reproducible. Continuously run.
Tests are authored as FHIR R4 TestScript Resources, run by an open Python executor against each server, and folded into a heatmap of servers × profiles. Each cell shows passed / total per MUST · SHOULD · MAY bucket. Read the methodology.
FHIR R4 base spec covers capability statement discovery, search semantics across every resource, error handling, terminology operations, history, and the Bulk Data kickoff surface. Bulk Data Access IG v2 tests the $export kickoff surface — Patient/$export request shape, 202 Accepted, Content-Location, capability declaration. US Core 6.1 and SMART on FHIR v2 columns are committed in shape and fill in as Inferno integration lands.
Each vendor has an independent scorecard with the exact passed / total counts, linked evidence, and an embeddable badge you can drop into a README: HAPI FHIR conformance · Medplum conformance · Aidbox conformance · Microsoft FHIR conformance · Blaze conformance · Spark conformance.
Existing FHIR conformance comparisons are run by FHIR vendors. mock.health doesn't ship a FHIR server. The heatmap, the methodology, and the runner are open — vendors can submit corrections via PR.