Model choice is a clinical systems decision
PHIGate helps digital-health teams compare open healthcare models, understand the deployment constraints, and prepare a defensible routing policy before protected health information ever reaches a model.
The model name is not the deployment
Healthcare teams need all three answers. PHIGate connects model-specific clinical evaluation, license and provenance, and deployment-path controls in one decision record.
PHIGate is healthcare-only by intent. That is a choice about what we build.
As of July 2026, we have not identified another model gateway that pairs physician-led, task-level evaluations with per-model compliance metadata. competitive scan · 2026-10-20
A focused catalog of healthcare-specific open and open-weight models. Verification status is recorded per field, never per row, and every field carries a source, a retrieval date, the exact configuration scope, a verification status, a reviewer, and a re-verification due date. Where we do not know, the entry says unknown. Browse the directory.
A physician-led, task-level evaluation protocol, published as a draft for review. We have not run it to completion. No model is scored, ranked, compared, or recommended anywhere on this site. Read the draft.
One endpoint that carries a compliance profile on the API key and checks requests against it before inference. Design goals and mechanisms are documented in How it works. None of it is live.
US digital-health founders, CTOs, and engineers building patient-facing or clinician-facing AI features: intake, messaging, summarization, documentation, navigation. If you are close enough to real data to be worried about it, this is being built for you.
PHI inputs are prohibited on every PHIGate surface, and we do not solicit PHI anywhere. There is no PHI tier today. The BAA tier is in build and opens only when the agreements and controls that make it lawful exist. We will not imply otherwise, here or anywhere.