What Is OpenEvidence? Who Can Sign Up, Pricing, and How It Works (2026)

Hello, this is Ryuta Hamamoto from TIMEWELL.

Most people who search for "OpenEvidence" want answers to a few plain questions before anything else. What is it, exactly? Can I sign up? Does it cost anything? Can I put patient details into it? So here are the short answers as of October 4, 2026, based on OpenEvidence's own Terms of Use (revised September 17, 2026) and its public announcements.

Question	Short answer (as of October 4, 2026)
What is it?	A medical AI for healthcare professionals that answers clinical questions with citations to peer-reviewed literature and guidelines
How widely is it used?	The company says a majority of U.S. physicians use it daily on average, and more than one million physicians use it worldwide
Who can sign up?	Physicians and other healthcare professionals. The terms say it is not intended for patients or consumers
Where is it unavailable?	The EU, the UK, the EEA, and countries under a U.S. trade embargo
What does it cost?	Free for verified U.S. clinicians, funded by advertising
Patient data?	U.S. covered entities can enter PHI under a BAA. Everyone else should follow local law and institutional policy

I have updated this article several times since January. This version replaces numbers that no longer hold. The old headline figure, "roughly 40% of U.S. physicians," came from OpenEvidence's October 2025 announcement. Its 2026 boilerplate now says "the majority," so I changed the article to match.

One note before going further. TIMEWELL is not a medical provider, and nothing here is medical advice. Clinical decisions belong to the treating clinician.

What OpenEvidence is, according to its own documents

OpenEvidence is built by OpenEvidence Inc., which runs the service from its offices in Miami, Florida¹. Founder and CEO Daniel Nadler previously built Kensho, the economic-data AI company that S&P Global acquired for roughly $550 million, and co-founder Zachary Ziegler came out of machine learning research at Harvard². In March 2023 the company joined the Mayo Clinic Platform accelerator³, which is where it started tightening the product against real clinical feedback.

The Terms of Use are blunt about what the product is not. The service is "intended for use by physicians and other healthcare professionals acting in their professional capacity," and the information is "provided for educational and informational purposes only." The terms go on to say it is not meant to serve as a diagnostic service, to recommend a particular product or therapy, or to substitute for a qualified clinician's judgment¹. "ChatGPT for doctors" is a catchy label, but the legal framing is closer to a research assistant with very good references.

Those references come from licensed content. The About page lists official partnerships with NEJM, JAMA and the JAMA Network specialty journals, NCCN guidelines, Nature and the Nature Portfolio, and Cochrane Systematic Reviews⁴. The NEJM deal was announced in February 2025⁵ and the JAMA Network deal in June 2025⁶. The list kept growing through 2026, with Springer Nature in August and Sage in September⁷.

Adoption is the part that changed most. In October 2025, OpenEvidence said more than 40% of U.S. physicians logged in daily on average⁸. The boilerplate attached to its October 1, 2026 announcement says the product is "now used daily, on average, by the majority of physicians in the United States"⁹. A September 11, 2026 release added that it is trusted by more than one million physicians worldwide, has supported over 500 million clinical consultations from verified professionals globally, and is used by more than 50% of physicians in Canada¹⁰. These are the company's own figures, so read them as such. Even with that discount, this is no longer a tool people try once and forget.

The business runs on advertising, not physician subscriptions. In January 2026 the company raised $250 million at a $12 billion valuation, with annual revenue reported above $100 million¹¹. In late September, Axios reported a further $250 million raise at a $15 billion valuation¹². I could not find an official announcement of that second round as of October 3, so treat it as reported, not confirmed.

Who can sign up, where it works, and what it costs

Eligibility is set in the Terms of Use, and the regional clause is the one most people miss. As of the September 17, 2026 revision, "the Services are not available in the European Union (EU), United Kingdom (UK), the European Economic Area (EEA), or any country subject to a U.S. trade embargo"¹. The same terms reserve the right to limit availability to any person or region at any time. If you practice outside the United States, check the current terms on the day you register. Europe and the UK are a reminder that availability can change.

Registration asks for your name, email, profession, specialty, and your National Provider Identifier (NPI) or, as applicable, another professional license number¹. For U.S. clinicians the NPI path is the standard one. Canadian physicians gained a dedicated route in September 2026: they can verify with their MINC number, the lifetime identifier held by more than 95% of physicians in Canada, which OpenEvidence says keeps their access free¹⁰. In Japan, GIGAZINE reported in June 2026 that the mobile app asked for an image of a certificate showing the physician's national registration number before it would create an account¹³, and other Japanese clinicians have reported getting verified with Japanese licenses¹⁴.

The steps themselves are short. Open openevidence.com or install the iOS or Android app, which has been available since August 2024¹⁵. Create an account, choose your profession and specialty, submit your license details, and start once verification clears. Verification is not always instant, so it is better done on a quiet afternoon than five minutes before clinic.

On pricing, the company's own language is precise: "OpenEvidence is free for all verified U.S. clinicians"⁹. Its advertising policy goes a step further and says the company is committed to providing the service "for free and equally to all health care professionals," from large institutions to small practices¹⁶. I did not find an official price list for other countries. Separately, Reuters reported on September 23, 2026 that Anthropic and OpenEvidence would offer a specialized version free of charge to clinicians in about 100 low- and middle-income countries, including Uganda, Angola, Sudan, Haiti, and Mongolia¹⁷. That initiative targets those countries specifically and does not change terms elsewhere.

One clause matters more than people expect. If you work for a health system, hospital, or other healthcare organization, the terms require that your use complies with that organization's policies, "including any restrictions on the use of artificial intelligence tools or third-party platforms"¹. Signing up as an individual does not override your employer's AI policy. For organizations in Japan that are still drafting one, we publish generative-AI adoption proposal and usage policy templates (free; company name and work email required). They are written for general companies under Japanese privacy rules, so a hospital would still need to add its own patient-data provisions.

How answers are built, and how to read them

Under the hood, OpenEvidence uses retrieval-augmented generation. For each question it pulls relevant passages from its licensed literature and guidelines, then writes an answer with citations you can open on the spot. In July 2026 it added EvidenceGrade, which grades the strength of the cited evidence using an approach built on the GRADE framework¹⁸. That targets a real weakness of AI summaries, which tend to blend a randomized trial and a small observational study into one confident paragraph. In September 2026 the company introduced a model family with different time budgets: Osler as the fast default at about five seconds, Sackett at about thirty seconds for questions that turn on the weight of the evidence, and Snow at about five minutes for a full literature investigation¹⁹. DeepConsult, launched in July 2025 for longer research reports, sits on the same path²⁰.

The company also cites exam results. In August 2025 it said it was the first AI to score 100% on the USMLE²¹. I treat that as a signal that the system handles medical text well, not as evidence that it can replace clinical judgment. Exams have clean questions. Patients do not.

Here is how I would read an answer, especially from outside the United States. First, check which country and which population the cited studies come from. Second, look for drugs or regimens that are standard in the U.S. but unapproved, or covered differently, where you practice. Third, compare the answer with your national or specialty society guidelines. OpenEvidence leans on U.S. and English-language sources, and its announcements list, as I read it on October 3, 2026, shows no partnership with a Japanese society or publisher⁷. Japanese users report that questions asked in Japanese come back in Japanese¹³, but answering in Japanese is not the same as answering for Japanese practice.

Comparison	OpenEvidence	General-purpose chatbots	UpToDate
Source material	Licensed medical literature and guidelines	Broad web text	Articles written and revised by specialists
Citations	Shown with every answer	Often missing or unreliable	References per article
How you use it	Ask a question, get a cited answer	Ask anything, with medical-use caveats	Search and read structured articles
Price for U.S. physicians	Free, ad-supported	Free and paid tiers	Paid subscription
Who can use it	Verified healthcare professionals	Anyone	Subscribing clinicians and institutions

Competition is also getting crowded. In April 2026 OpenAI launched ChatGPT for Clinicians, free for verified U.S. physicians, nurse practitioners, physician assistants, and pharmacists²². My own view is that the two kinds of tools work best together: a quick cited-answer tool between patients, and a curated reference when you need to relearn a whole topic. Whatever the tool, the habit that matters is opening the sources and reading them yourself.

Patient data, HIPAA, and the rules outside the U.S.

Start with HIPAA, because it is the most misunderstood piece. On April 25, 2025, OpenEvidence announced that it complies with HIPAA requirements and that U.S. covered entities can enter protected health information, handled under the company's Business Associate Agreement²³. That is a real step for U.S. hospitals and practices. It is not a general permission to paste chart notes into any account.

The privacy policy says raw user queries are confidential and that the company does not train AI models on protected health information. It also says personal information may be transferred to the United States or other jurisdictions whose privacy laws may be less protective than yours²⁴. The Terms of Use put the burden on you: if you submit information that could identify a patient, you confirm you have the necessary consents and authorizations, and you are responsible for any failure to de-identify¹. My default advice is simple. If the clinical question can be asked without identifiers, ask it without identifiers.

Enterprise deployments work differently from individual accounts. When Cedars-Sinai integrated OpenEvidence with Epic in May 2026, the company stated that patient information from the EHR would be used only to support care decisions for that patient and would not be stored after the clinical session²⁵. Those terms come from a contract between a health system and a vendor. A free individual account accepted with a click is a different arrangement. The same logic applies to Visits, OpenEvidence's ambient documentation feature, which the company describes as free for verified U.S. healthcare professionals²⁶. Recording patient conversations into any third-party service is an institutional decision, not a personal experiment.

For readers in Japan, the domestic rules are specific. The Personal Information Protection Commission's June 2, 2023 alert says businesses entering personal information into generative AI services should make sure it stays within the specified purpose of use, and that entering personal data without consent may violate the law if the data is handled for purposes other than producing the response²⁷. Hospitals also fall under the Ministry of Health, Labour and Welfare's security guidelines for medical information systems, revised to version 7.0 in June 2026²⁸. The September 2026 Q&A for those guidelines (question 18) says that when medical information is entered into a generative AI service, the server does not need to be under Japanese jurisdiction if a contract guarantees the information will not be stored²⁹. Again, that assumes a contract, which an individual sign-up does not provide.

Regulatory status is a separate question. Software intended for diagnosis or treatment can be regulated as a medical device, including in Japan, where it falls under the program medical device (SaMD) framework³⁰. OpenEvidence describes itself in its terms as an educational and informational service, not a diagnostic one, and I have not seen an announcement of Japanese medical-device approval. Japan's Healthcare AI Platform Collaborative Innovation Partnership (HAIP) also publishes generative AI guidelines for healthcare, with the second edition dated July 11, 2025³¹. Across all of these documents, one principle stays fixed: the clinician makes the final call.

What changed since the last version of this article

A lot moved in the three months after my July update. Here is the before-and-after in one place.

Item	Previous version	As of October 2026
U.S. adoption	About 40% of U.S. physicians daily	A majority of U.S. physicians daily on average, per company boilerplate9
Global use	No official figure cited	More than one million physicians and 500 million+ consultations globally (September 11, 2026)10
Availability	Regional limits not covered	Not available in the EU, UK, or EEA per the Terms of Use; Canada added MINC verification with free access110
Valuation	$12B (January 2026)	$15B round reported in September 2026, not officially confirmed12
Health-system partners	EHR integrations at Mount Sinai, Sutter Health, and Cedars-Sinai	NewYork-Presbyterian (July), Penn Medicine, MSK, and UTMB (September), AdventHealth (October)7
Product	DeepConsult	EvidenceGrade (July), Patient Take-Homes for physician-curated handouts (August), the Osler, Sackett, and Snow models (September)183219
Competition	Healthcare versions of general AI products	OpenAI's free ChatGPT for Clinicians for verified U.S. clinicians (April 2026)22

I also removed several third-party estimates from earlier versions, such as monthly sign-ups and revenue per user, because I could not confirm them against primary sources. And I corrected one detail: earlier versions said Mount Sinai's deployment covered seven hospitals, but Mount Sinai's official announcement describes a health system of six hospitals³³. Sutter Health's announcement is dated February 11, 2026³⁴.

Two shifts stand out to me. The product is moving from a literature search box to an agent that reads the patient's chart alongside the evidence. And availability is splitting by region. The service is not offered in Europe, Canada now has a national verification route, and low- and middle-income countries are getting a free version through a partnership. For clinicians outside the U.S., access now depends on a company decision that can change with a terms update.

What enterprise AI teams can borrow from this design

Strip away the clinical packaging and five design choices remain. Narrow the corpus to sources with real quality control. Attach citations so a person can verify the original. Verify who is using the system before opening deeper features. Partner with the institutions that own the source of truth. And, new in 2026, show how strong the underlying evidence is, not just what it says.

Those choices transfer well beyond medicine. Law firms working from case law, manufacturers working from technical standards and patents, and financial institutions working from regulatory text all face the same three blockers I hear in nearly every enterprise AI project. Pasting confidential files into a general chat tool collides with NDAs and security policy. General models do not reliably ground answers in your own documents. And answers without sources stall at the legal or board review stage. OpenEvidence did not solve hallucination. It made checking the answer part of normal use, which is a more honest design.

The advertising model deserves a second look too, especially for pharmaceutical companies. OpenEvidence's advertising policy says the answer system and the ad system are fully separate and that advertisers cannot influence answers¹⁶. Whether users keep believing that line will decide whether the product stays trusted infrastructure. Any company that uses AI to deliver medical information should decide early how evidence and promotion are separated, and who reviews that separation.

If you are not a clinician, you probably cannot sign up, and that is the point of the design. But it is worth studying once as a working example of a domain corpus, citations, verified users, and institutional partners. If you run a Japan-based hospital, pharmaceutical company, or enterprise team and want help turning that pattern into an internal AI policy and rollout plan, get in touch. For a broader view of how AI is being used in diagnostics, drug discovery, and medical administration, see AI in healthcare: implementation patterns.

Summary

OpenEvidence is a cited-answer medical AI for verified healthcare professionals, and by its own October 2026 account a majority of U.S. physicians use it daily. U.S. clinicians can sign up with an NPI and use it free. The September 2026 terms exclude the EU, the UK, and the EEA, while Canada now has a national verification route and clinicians in other countries, including Japan, report using it with local licenses. HIPAA coverage is a contract-based arrangement for U.S. covered entities, not a blanket permission, and outside the U.S. local privacy law and institutional policy decide what you can enter.

If you are about to sign up, read your employer's AI policy and the current OpenEvidence terms side by side first. Ten minutes there will save you from an awkward conversation about patient data later.

Related Articles

• AI in Healthcare: Diagnostic Support, Drug Discovery, and Medical Administration
• Enterprise AI Agent Comparison: Selection Criteria and Major Products
• Enterprise AI Glossary: Key Concepts You Should Know

References

Footnotes

1. OpenEvidence Terms of Use (intended users, educational and informational purpose, EU/UK/EEA and embargo exclusion, NPI or other license number at registration, compliance with employer AI policies, responsibility for identifiable information) — OpenEvidence — last updated 2026-09-17 — https://www.openevidence.com/policies/terms ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷

2. OpenEvidence Business Breakdown & Founding Story (founders' background) — Contrary Research — accessed 2026-07 — https://research.contrary.com/company/openevidence ↩

3. OpenEvidence to Become a Mayo Clinic Platform Accelerate Company — OpenEvidence announcement — 2023-03-28 — https://www.openevidence.com/announcements/openevidence-to-become-a-mayo-clinic-platform-accelerate-company ↩

4. About (official partnerships with NEJM, the JAMA Network, NCCN, Nature Portfolio, and Cochrane) — OpenEvidence — accessed 2026-10-03 — https://www.openevidence.com/about ↩

5. OpenEvidence Achieves $1 Billion Valuation in Sequoia-led Round and Announces Content Partnership with the New England Journal of Medicine (official OpenEvidence news release via PR Newswire) — 2025-02-19 — https://www.prnewswire.com/news-releases/openevidence-achieves-1-billion-valuation-in-sequoia-led-round-and-announces-content-partnership-with-the-new-england-journal-of-medicine-302380960.html ↩

6. OpenEvidence and the JAMA Network Sign Strategic Content Agreement (official OpenEvidence news release via PR Newswire) — 2025-06-05 — https://www.prnewswire.com/news-releases/openevidence-and-the-jama-network-sign-strategic-content-agreement-302473690.html ↩

7. Announcements (Springer Nature 2026-08-05, Sage 2026-09-29, NewYork-Presbyterian 2026-07-09, Penn Medicine 2026-09-15, MSK 2026-09-16, UTMB 2026-09-22, AdventHealth 2026-10-01, and others) — OpenEvidence — accessed 2026-10-03 — https://www.openevidence.com/announcements ↩ ↩² ↩³

8. OpenEvidence Collaborates with Microsoft to Expand AI Leadership in Healthcare (40%+ of U.S. physicians log in daily on average) — OpenEvidence announcement — 2025-10 — https://www.openevidence.com/announcements/openevidence-collaborates-with-microsoft-to-expand-ai-leadership-in-healthcare-bringing-clinical-evidence-and-guidelines-to-enterprise-clinician-workflows ↩

9. AdventHealth and OpenEvidence Together Support More Time for Patient Care (boilerplate: used daily by the majority of U.S. physicians; free for all verified U.S. clinicians) — OpenEvidence announcement — 2026-10-01 — https://www.openevidence.com/announcements/adventhealth-and-openevidence-together-support-more-time-for-patient-care ↩ ↩² ↩³

10. OpenEvidence and MINC#NIMC Partner to Ensure Free Access for Canadian Physicians to Leading Medical AI Platform (one million+ physicians worldwide, 500 million+ consultations, 50%+ of Canadian physicians) — OpenEvidence announcement — 2026-09-11 — https://www.openevidence.com/announcements/openevidence-and-mincnimc-partner-to-ensure-free-access-for-canadian-physicians-to-leading-medical-ai-platform ↩ ↩² ↩³ ↩⁴

11. OpenEvidence hits $12B valuation with new round led by Thrive, DST — TechCrunch — 2026-01-21 — https://techcrunch.com/2026/01/21/openevidence-hits-12b-valuation-with-new-round-led-by-thrive-dst/ ↩

12. OpenEvidence reportedly raises $250M at $15B valuation — Axios Pro — 2026-09-25 — https://www.axios.com/pro/health-tech-deals/2026/09/25/openevidence-250m-raise-15b-valuation-a16z ↩ ↩²

13. We asked OpenEvidence, an AI built for healthcare professionals, about long-standing symptoms (Japanese-language questions and answers; license number check at app registration) — GIGAZINE — 2026-06-01 — https://gigazine.net/news/20260601-openevidence-medical-ai-review/ ↩ ↩²

14. What is OpenEvidence? (report on use from Japan) — decades.co.jp — 2026-02 — https://decades.co.jp/openevidence-clinical-guide-202602/ ↩

15. OpenEvidence is now available for iOS and Android — OpenEvidence announcement — 2024-08-20 — https://www.openevidence.com/announcements/openevidence-is-now-available-for-ios-and-android ↩

16. OpenEvidence Advertising Policy (free and equal access for healthcare professionals, separation of ads and answers) — OpenEvidence — last updated 2024-08-12 — https://www.openevidence.com/policies/advertising ↩ ↩²

17. Exclusive: Anthropic, OpenEvidence partner to bring medical AI worldwide (Reuters; a correction was issued after publication) — The Star (Reuters) — 2026-09-23 — https://www.thestar.com.my/tech/tech-news/2026/09/23/exclusive-anthropic-openevidence-partner-to-bring-medical-ai-worldwide ↩

18. OpenEvidence Launches EvidenceGrade — OpenEvidence announcement — 2026-07-10 — https://www.openevidence.com/announcements/openevidence-launches-evidencegrade-empowering-physicians-to-see-the-strength-of-cited-evidence-beneath-each-ai-answer ↩ ↩²

19. Introducing the OpenEvidence Model Family (Osler, Sackett, Snow) — OpenEvidence announcement — 2026-09-03 — https://www.openevidence.com/announcements/introducing-the-openevidence-model-family ↩ ↩²

20. OpenEvidence announces $210 million round at $3.5 billion valuation (Series B and DeepConsult announcement) — PR Newswire — 2025-07 — https://www.prnewswire.com/news-releases/openevidence-the-fastest-growing-application-for-physicians-in-history-announces-210-million-round-at-3-5-billion-valuation-302505806.html ↩

21. OpenEvidence Creates the First AI in History to Score a Perfect 100% on the United States Medical Licensing Examination (USMLE) (official OpenEvidence news release via PR Newswire) — 2025-08-15 — https://www.prnewswire.com/news-releases/openevidence-creates-the-first-ai-in-history-to-score-a-perfect-100-on-the-united-states-medical-licensing-examination-usmle-302531156.html ↩

22. OpenAI launches ChatGPT for Clinicians to expand AI support for U.S. healthcare professionals — HLTH — 2026-04-24 — https://hlth.com/insights/news/openai-launches-chatgpt-for-clinicians-to-expand-ai-support-for-u-s-healthcare-professionals ↩ ↩²

23. OpenEvidence is now HIPAA compliant (U.S. covered entities may enter PHI under the BAA) — OpenEvidence announcement — 2025-04-25 — https://www.openevidence.com/announcements/openevidence-is-now-hipaa-compliant ↩

24. OpenEvidence Privacy Policy: Medical Professionals (confidential queries, no training on PHI, international data transfer) — OpenEvidence — last updated 2026-09-17 — https://www.openevidence.com/policies/privacy ↩

25. OpenEvidence Partners with Cedars-Sinai to Create Patient-Aware Clinical Intelligence With Agentic Clinical AI (EHR patient information not stored after the clinical session) — OpenEvidence announcement — 2026-05-20 — https://www.openevidence.com/announcements/openevidence-partners-with-cedars-sinai-to-create-patient-aware-clinical-intelligence-with-agentic-clinical-ai ↩

26. OpenEvidence Launches "Visits" (free for verified U.S. healthcare professionals) — OpenEvidence announcement — 2025-08-21 — https://www.openevidence.com/announcements/visits-real-time-medical-intelligence ↩

27. Alert on the use of generative AI services — Personal Information Protection Commission, Japan — 2023-06-02 — https://www.ppc.go.jp/news/careful_information/230602_AI_utilize_alert/ ↩

28. Guidelines for the Security Management of Medical Information Systems, Version 7.0 — Ministry of Health, Labour and Welfare, Japan — 2026-06 — https://www.mhlw.go.jp/stf/shingi/0000516275_00006.html ↩

29. Q&A on the Guidelines for the Security Management of Medical Information Systems (question 18, entering medical information into generative AI services) — Ministry of Health, Labour and Welfare, Japan — 2026-09 — https://www.mhlw.go.jp/content/10808000/001752355.pdf ↩

30. Program medical devices — Ministry of Health, Labour and Welfare, Japan — accessed 2026-10-03 — https://www.mhlw.go.jp/stf/seisakunitsuite/bunya/0000179749_00004.html ↩

31. Generative AI Use Guidelines in the Medical and Healthcare Field, 2nd edition — Healthcare AI Platform Collaborative Innovation Partnership (HAIP) — 2025-07-11 — https://haip-cip.org/assets/documents/nr_20241002_02.pdf ↩

32. OpenEvidence Launches Patient Take-Homes (optional physician-curated handouts; patients cannot ask their own questions) — OpenEvidence announcement — 2026-08-19 — https://www.openevidence.com/announcements/openevidence-launches-patient-take-homes ↩

33. Mount Sinai Health System Collaborates with OpenEvidence to Provide Evidence-Based Knowledge Within Electronic Medical Record — OpenEvidence announcement — 2026-03-31 — https://www.openevidence.com/announcements/mount-sinai-health-system-collaborates-with-openevidence-to-provide-evidence-based-knowledge-within-electronic-medical-record ↩

34. Sutter Health Collaborates with OpenEvidence to Bring Evidence-Based, AI-Powered Insights into Physician Workflows — OpenEvidence announcement — 2026-02-11 — https://www.openevidence.com/announcements/sutter-health-collaborates-with-openevidence-to-bring-evidence-based-ai-powered-insights-into-physician-workflows ↩