ChatGPT for Clinicians: who qualifies, what it does and what to use outside the US (2026)

OpenAI describes ChatGPT for Clinicians as a version of ChatGPT for verified clinicians in the United States, designed to support clinical work including evidence review, documentation and medical research ^[1]. The announcement on 22 April 2026 made it free for any verified physician, NP, PA or pharmacist, starting in the US ^[2]. OpenAI tied the launch to clinician demand, citing a 2026 American Medical Association survey in which 72% of physicians reported using AI in clinical practice, up from 48% the year before ^[2].

ChatGPT for Clinicians is a workspace, not a separate app. After verification a clinician gets a new ChatGPT for Clinicians workspace, keeps their existing ChatGPT workspace, and switches between the two from the account switcher ^[1]. At launch OpenAI said ChatGPT for Clinicians builds on models such as GPT-5.4 ^[2]; the chatgpt.com plan page now advertises higher limits on GPT-6 for healthcare work ^[3].

OpenAI's own pre-release evaluation reports that physician advisors tested 6,924 conversations from their daily work and rated 99.6% of responses safe and accurate, and that advisors have reviewed more than 700,000 model responses overall ^[2]. These are OpenAI's figures, not an independent study, and OpenAI states that ChatGPT for Clinicians is designed to support clinicians, not replace their judgment ^[2].

The two OpenAI lists differ. The help center, updated in October 2026, names four professions ^[1]; the chatgpt.com plan page says verified US-based licensed clinicians including physicians, NPs, PAs, psychologists and other licensed clinicians ^[3]. A registered nurse or another licensed professional who is not on the help-center list should expect the verification step, not the marketing page, to decide.

The NPI requirement is the hard edge. The National Provider Identifier is a US identifier, so a physician licensed in Canada, the UK, the EU, Brazil, Mexico or India without a US NPI cannot complete sign-up. EvidenceMD runs no NPI or licence check, so the same clinician can start using it the day they register ^[5].

OpenAI's announcement says ChatGPT for Clinicians is available for free to verified individual clinicians in the US ^[2], and the help center answers "Is ChatGPT for Clinicians free?" with "At launch, ChatGPT for Clinicians is free for verified clinicians in the United States" ^[1]. OpenAI has published no paid tier or end date for the free access.

Organisations are priced differently. ChatGPT for Healthcare, the version for hospitals and health systems, is priced on the ChatGPT Enterprise model: the organisation buys a shared credit pool sized to the deployment, and access starts with OpenAI's sales team ^[4].

The common use cases OpenAI lists are reviewing care pathways, synthesising evidence with citations, reasoning through differentials, and drafting referral letters, prior authorisations and plain-language patient instructions ^[1]. That is mostly the work around a clinical decision. For the decision itself, OpenAI's wording is explicit: ChatGPT for Clinicians is intended to support, not replace, professional judgment, and clinicians remain responsible for all care decisions ^[1].

Where EvidenceMD differs is in what it shows. EvidenceMD is built for clinical reasoning: it returns a ranked differential diagnosis and an assessment and plan, and streams its full reasoning with a citation bound to every claim, so a clinician can check how the answer was reached, not only where it came from ^[6][9].

Content shared with ChatGPT for Clinicians is not used to train OpenAI's models ^[1]. That is a training setting, not a compliance agreement: HIPAA coverage starts only when the BAA is signed ^[1][7]. Outside the US the BAA question does not arise, because the product is not available, and the local data-protection law applies instead ^[7].

EvidenceMD offers an individual BAA too. It applies to EvidenceMD's Free, Individual and Max plans and is accepted electronically; Enterprise customers sign a separately negotiated agreement ^[8].

OpenAI's help center answers the question directly: at launch ChatGPT for Clinicians is available to verified clinicians in the United States, and OpenAI plans to expand access to additional countries over time ^[1]. The April 2026 announcement adds the first step: in the coming months OpenAI will work with the Better Evidence Network to pilot access for verified clinicians outside the United States, as permitted by local regulations ^[2].

Until that happens, a clinician in Canada, the UK, the EU, Latin America, Brazil, India or elsewhere has three practical routes. Consumer ChatGPT is available, but it carries no clinician verification, no clinical-search workspace and no BAA ^[7]. OpenEvidence is also US-focused: its standard sign-up asks for a US NPI ^[10]. EvidenceMD works in every country with no NPI and answers in 30 languages, including Spanish and Portuguese ^[5].

The useful split is by job. For drafting letters, notes and patient instructions inside a US practice, ChatGPT for Clinicians is a strong free option with a BAA available ^[1]. For a fast cited lookup in the US, OpenEvidence is free with an NPI ^[10]. For reasoning through a case anywhere in the world, with the reasoning on screen and a citation for every claim, EvidenceMD is built for that job and free to start ^[5][6].

For most clinicians outside the US the deciding questions are the same ones OpenAI's launch raises: can I sign up, can I check the sources, and where does my patient data go. EvidenceMD answers the first two directly: registration needs no NPI, and every claim carries a citation you can open ^[5][6]. On the third, de-identify patient details unless your local rules and your agreement with the vendor allow otherwise ^[7].