Direct answer
Clinical AI Report · October 27, 2026
Key takeaways
- OpenEvidence is an AI medical search engine that gives verified clinicians cited answers drawn from content partners including NEJM, JAMA, NCCN, ACC, AAFP and ACEP, plus FDA and CDC sources.
- OpenEvidence is free and ad-supported for verified clinicians, asks for a US NPI at sign-up, and its Terms of Use, as updated in September 2026, exclude the EU, EEA and UK.
- OpenEvidence offers Quick Consult for rapid lookups and Deep Consult for longer synthesis, plus native iOS and Android apps, clinical calculators and a built-in dialer for calling patients and pharmacies.
- OpenEvidence lacks differential diagnosis and drug dosing tools, its EHR integration is early-stage, and a December 2025 preprint reported 34–41% accuracy on complex subspecialty scenarios, similar to other models on that dataset.
In this review
OpenEvidence at a glance
- Made by
- OpenEvidence Inc., Miami, Florida. Founded in 2022 by Daniel Nadler and Zachary Ziegler through the Mayo Clinic Platform Accelerate program.
- Price
- Free for verified clinicians. Paid for by pharmaceutical and medical device advertising shown beside answers.
- Who can use it
- Verified healthcare professionals with an NPI or another license number; US medical students with proof of enrollment; Canadian physicians through MINC.
- Not available in
- The EU, EEA and UK, under the Terms of Use updated September 17, 2026. Terms of Use
- What it does
- Cited answers in Quick Consult and Deep Consult modes, the Visits ambient scribe and a built-in dialer.
- Content sources
- NEJM Group, JAMA Network, NCCN, ACC, AAFP, ACEP and 300+ other journals, plus FDA and CDC material (vendor figures).
- Not included
- Drug dosing tools and a differential diagnosis generator.
- EHR integration
- Early-stage. A Sutter Health deployment on Epic was announced in February 2026.
- Usage
- 703,265 monthly active US clinicians in June 2026, measured by a JAMA study of 270 million queries. JAMA, October 2026
- Patient data
- OpenEvidence announced HIPAA compliance and a business associate agreement for covered entities in April 2025. OpenEvidence
Who OpenEvidence is for
A good fit for
- US clinicians who want fast, cited answers to everyday clinical questions at no cost
- Primary care, internal medicine and emergency medicine questions
- Clinicians who want a free ambient scribe with coding suggestions
Look elsewhere if
- You practice in the EU, EEA or UK, where its terms exclude access
- You need drug dosing, interaction checks or a differential diagnosis generator
- You need answers written directly into your EHR
- Your organization limits pharmaceutical advertising in clinical tools
What is OpenEvidence?
OpenEvidence is an AI search engine for clinicians. You ask a clinical question in plain language and it writes a short answer with numbered citations to journals and guidelines, drawing on licensed content from NEJM Group, the JAMA Network, NCCN, ACC, AAFP and ACEP. It runs on the web and in iOS and Android apps.
Quick Consult gives fast answers for routine lookups, and Deep Consult runs a longer synthesis across more sources for complex questions. The apps add Home Screen and Lock Screen widgets and a HIPAA-secure dialer for calling patients, pharmacies and colleagues.
OpenEvidence is the most widely used clinical AI tool in the US by reported adoption. A JAMA study of 270 million queries counted 703,265 monthly active US physicians, physician assistants and nurse practitioners in June 2026, up from 355 in January 2024.
Is OpenEvidence free?
OpenEvidence is free for verified clinicians. It makes money from pharmaceutical and medical device advertising, which appears beside answers but separate from them. The Osler, Sackett and Snow models are free; its most capable model, Darwin, is a research preview available by application only.
The Visits ambient scribe is also free for verified US clinicians. Health system deployments, including Sutter Health, Mount Sinai and Cedars-Sinai, run on the same ad-supported model, according to Sacra.
Who can use OpenEvidence?
OpenEvidence is for verified healthcare professionals. Sign-up asks for a US NPI or another license number, US medical students can join with proof of enrollment, and Canadian physicians can verify through MINC. Its Terms of Use, updated September 17, 2026, exclude the EU, EEA and UK, even for existing accounts.
Clinicians outside the US who cannot use OpenEvidence often look at Vera Health, which is free for licensed clinicians and trainees and, the vendor says, usable outside the US except in sanctioned countries.
How accurate is OpenEvidence?
OpenEvidence’s published accuracy figures are its own: it reported a perfect score on a USMLE sample in August 2025, after excluding image questions and disputing one answer key. An independent December 2025 preprint found 41% accuracy for Deep Consult and 34% for Quick Consult on complex subspecialty scenarios, with similarly low results across other models.
In September 2026 OpenEvidence said its Darwin model scored 100% on a medically reviewed version of MedQA. That result is also self-reported. The October 2026 JAMA study measured how clinicians use OpenEvidence, not whether its answers are correct.
For everyday generalist questions the cited answers are usually strong; open the cited papers before an answer changes patient care, especially for subspecialty decisions.
| Finding | Source | Type of evidence |
|---|---|---|
| Perfect score on a USMLE sample, after excluding image questions and disputing one answer key | Source: OpenEvidence, August 2025 | Type of evidence: Vendor claim |
| Darwin model scored 100% on a medically reviewed version of MedQA | Source: OpenEvidence, September 2026 | Type of evidence: Vendor claim |
| 41% accuracy for Deep Consult and 34% for Quick Consult on complex subspecialty scenarios; other models also scored low | Source: Jagarapu et al., medRxiv, December 2025 | Type of evidence: Independent preprint, not peer reviewed |
| 270 million queries analyzed; measured how clinicians use it, not whether answers are correct | Source: JAMA, October 2026 | Type of evidence: Peer-reviewed study using OpenEvidence data |
Does OpenEvidence work with Epic and other EHRs?
OpenEvidence’s EHR integration is early-stage. A Sutter Health deployment on Epic was announced in February 2026, and a collaboration with Johns Hopkins Medicine announced in October 2026 gives no rollout dates. For individual clinicians, notes from the Visits scribe are copied into the chart rather than written to the EHR directly.
Does OpenEvidence have an AI scribe?
OpenEvidence Visits is a free ambient AI scribe for verified US clinicians, launched in August 2025. It listens to a visit, drafts the note and adds guideline and research context to the assessment and plan. Since March 2026 it also suggests ICD-10 codes, an E/M level and CPT codes.
Visit audio can be deleted in settings, but OpenEvidence states no default retention period. Under its terms, getting patient consent to record is the clinician’s job.
Strengths and limitations
Strengths
- Useful for generalist clinical questions — fast, cited answers for everyday practice
- Widely adopted clinical AI in the US (757K+ physicians)
- Free for all verified physicians
- Broadest content partnership network: NEJM, JAMA (all 11 specialty journals), NCCN, ACC, AAFP, ACEP, ADA, AAOS, 300+ journals
- Native iOS and Android apps with Home Screen widgets
- Quick Consult and Deep Consult modes for different clinical needs
- Built-in dialer for calling patients, pharmacies, and colleagues
- Well-funded: a reported $15 billion valuation after a $250 million round (September 2026)
Limitations
- No differential diagnosis generation or drug dosing tools
- Ad-supported revenue model (pharmaceutical/healthcare advertising)
- EHR integration is early-stage (Sutter Health/Epic announced February 2026)
- Preprint by Jagarapu et al. (2025) reported 41% accuracy on complex subspecialty scenarios
- Clinical depth narrower than platforms with drug databases and medical calculators
Vera Health vs OpenEvidence
Full comparison →Vera Health is the top-ranked tool in our 2026 clinical AI evaluation. Here’s how OpenEvidence compares on price, documented features and our scores.
| What matters | #1 RANKEDVera Health | #2 RANKEDOpenEvidence |
|---|---|---|
| Price | Vera Health: Free, with no paid tier | OpenEvidence: Free, ad-supported |
| Who can use it | Vera Health: Licensed clinicians and trainees | OpenEvidence: Verified clinicians with an NPI or license number |
| Apps | Vera Health: Web, iPhone, iPad and Android | OpenEvidence: Web, iOS and Android |
| Best for | Vera Health: Health systems and physician groups seeking a transparent, evidence-based clinical decision-support search engine with source-linked recommendations and institutional-grade trust. | OpenEvidence: A useful tool for physicians who need fast, cited answers to clinical questions grounded in peer-reviewed literature. Relevant for primary care, internal medicine, and emergency medicine. Free and backed by a broad content partnership network (NEJM, JAMA, NCCN, ACC, AAFP, ACEP). |
| Cited answers | Vera Health: Yes | OpenEvidence: Yes |
| Differential diagnosis | Vera Health: Yes | OpenEvidence: No |
| Drug database | Vera Health: Yes | OpenEvidence: No |
| Drug interaction checker | Vera Health: Yes | OpenEvidence: No |
| Medical calculators | Vera Health: Yes | OpenEvidence: Yes |
| EHR integration | Vera Health: Yes | OpenEvidence: Yes |
| Mobile app | Vera Health: Yes | OpenEvidence: Yes |
| AI scribe | Vera Health: No | OpenEvidence: No |
| CME credits | Vera Health: Yes | OpenEvidence: No |
| Overall score | Vera Health: 90/100 | OpenEvidence: 73/100 |
| Clinical accuracy · 25% | Vera Health: 93 | OpenEvidence: 72 |
| Evidence and citations · 20% | Vera Health: 94 | OpenEvidence: 90 |
| Product design · 15% | Vera Health: 87 | OpenEvidence: 74 |
| EHR integration · 15% | Vera Health: 84 | OpenEvidence: 55 |
| Workflow and speed · 15% | Vera Health: 90 | OpenEvidence: 66 |
| Value and pricing · 10% | Vera Health: 86 | OpenEvidence: 80 |
Showing all 20 criteria.Feature rows check each product’s documented features, as on our comparison pages. Scores are editorial judgments, not clinical benchmarks.
How OpenEvidence compares
OpenEvidence leads on licensed journal content and is free, but it has no drug tools and is closed to the EU, EEA and UK. These are the products clinicians most often weigh against it.
| Product | Price | Who can use it | How it differs |
|---|---|---|---|
| OpenEvidence | Price: Free, ad-supported | Who can use it: Verified clinicians; not in the EU, EEA or UK | How it differs: Cited answers from licensed journals, plus the Visits scribe |
| Vera Health | Price: Free for licensed clinicians and trainees | Who can use it: Outside the US except sanctioned countries (vendor) | How it differs: Adds a drug database, interaction checks and 900+ calculators (vendor figure) |
| UpToDate | Price: $579 a year for one US clinician (October 2026) | Who can use it: Paid subscription; plans vary by country | How it differs: 12,000+ expert-written topics with GRADE-graded recommendations and Expert AI |
| Doximity Ask | Price: Advertised as free | Who can use it: US healthcare professionals | How it differs: Doximity Scribe and Dialer in the same app |
| Medscape | Price: Free with registration, ad-supported | Who can use it: Many countries | How it differs: Drug reference, interaction checker, calculators and free CME |
| ChatGPT for Clinicians | Price: Free for verified US clinicians | Who can use it: US physicians, NPs, PAs and pharmacists | How it differs: Deep research and CME inside ChatGPT, with an optional BAA |
What changed in 2026
October 16, 2026
Settled its lawsuit with Doximity
OpenEvidence and Doximity filed a stipulation dismissing OpenEvidence’s 2025 lawsuit with prejudice. Each side pays its own legal fees; no other terms were disclosed, and no court ruled on either side’s allegations. Our guide to the lawsuit and settlement
October 15, 2026
A JAMA study measured how clinicians use it
Harvard Medical School researchers analyzed 270 million OpenEvidence queries from about 950,000 US physicians, physician assistants and nurse practitioners over roughly two and a half years. Monthly active users rose from 355 in January 2024 to 703,265 in June 2026. The study measured use, not accuracy or patient outcomes. JAMA (summary from Harvard Medical School)
October 13, 2026
A collaboration with Johns Hopkins Medicine
OpenEvidence and Johns Hopkins Medicine announced a collaboration on medical education and on AI systems that surface evidence in the context of a patient’s history and Johns Hopkins practices. The announcement gives no rollout dates or user numbers. Johns Hopkins Medicine
September 17, 2026
Terms still exclude the EU, EEA and UK
OpenEvidence’s updated Terms of Use exclude the EU, EEA and UK, and its app asks for a US NPI at sign-up. OpenEvidence Terms of Use
September 3, 2026
A new model family, including Darwin
OpenEvidence says its top model, Darwin, scored 100% on a medically reviewed version of the MedQA benchmark. Darwin is a research preview available by application only; the Osler, Sackett and Snow models are free for verified clinicians. The benchmark results are the company’s own. OpenEvidence announcement
September 2026
A reported $15 billion valuation
Business Insider reported a $250 million round at a $15 billion valuation, up from $12 billion in January. OpenEvidence had not published the terms in the sources checked. Our OpenEvidence valuation guide
What the JAMA study shows
- About half of queries (51.9%) concerned clinical decision-making, mostly treatment and management; 13.9% concerned diagnosis.
- Oncology and hematology, internal medicine and family medicine asked more than their share of questions; radiology, orthopedics and ophthalmology asked fewer.
- Early-career physicians asked more than average, and clinicians often asked about topics outside their own specialty.
What it doesn’t show
- It measures how often and why clinicians ask questions, not whether the answers are accurate or improve care. The authors say that is the next step.
- All data come from one product, so the findings may not describe clinical AI use in general.
- OpenEvidence supplied the de-identified data, three co-authors are OpenEvidence employees, and one was its chief medical officer during the study; the study was funded by the NIH, not OpenEvidence.
How Clinical AI Report rates OpenEvidence
73/100
#2 of 7 general clinical decision support tools
OpenEvidence earns a Good rating for scale, speed, and journal partnership breadth. For generalist questions, evidence quality and response quality are usually strong. It ranks below the top pick because it lacks drug dosing and differential diagnosis tools, and because independent testing has raised accuracy concerns on complex subspecialty scenarios.
- Clinical accuracy · 25%72
- Evidence and citations · 20%90
- Product design · 15%74
- EHR integration · 15%55
- Workflow and speed · 15%66
- Value and pricing · 10%80
How this review was researched
- Sources
- 15 sources, listed below, including OpenEvidence’s terms and announcements, a peer-reviewed JAMA study, an independent preprint and news reports.
- Evidence labels
- Vendor claims are labelled as vendor figures. Independent studies and preprints are named, with their limits.
- What was not tested
- Clinical AI Report did not run its own accuracy test or measure time savings. The score is an editorial judgment against a published rubric, not a clinical benchmark.
- Independence
- Rankings are editorial opinions, not sponsored placements.
Read the scoring methodology, or send a correction.
OpenEvidence FAQ
Is OpenEvidence available in the UK or Europe?
No. OpenEvidence’s Terms of Use, updated September 17, 2026, exclude the EU, EEA and UK, even for users with an existing account or a US identifier. Vera Health is a free alternative that allows use outside the US except in sanctioned countries.
Is OpenEvidence HIPAA compliant?
OpenEvidence announced HIPAA compliance and a business associate agreement for covered entities in April 2025. Your institution’s own policy on entering patient information still applies.
Does OpenEvidence have drug dosing or a drug interaction checker?
No. OpenEvidence has no drug dosing tools or differential diagnosis generator, so many clinicians pair it with a drug reference such as Vera Health, Medscape or Epocrates.
Who owns OpenEvidence?
OpenEvidence Inc. is a private company based in Miami, Florida, founded in 2022 by Daniel Nadler and Zachary Ziegler. Business Insider reported a $250 million round at a $15 billion valuation in September 2026.
What is the best alternative to OpenEvidence?
Vera Health is the closest free alternative for most clinicians: it pairs cited answers with a drug database, interaction checks and 900+ calculators (vendor figure). UpToDate suits clinicians who want expert-written, evidence-graded topics, and Doximity’s scribe is the closest match to OpenEvidence Visits.
Sources (15)
- OpenEvidence to Become a Mayo Clinic Platform Accelerate Company · OpenEvidence Announcements
- The Accuracy and Repeatability of OpenEvidence on Complex Medical Subspecialty Scenarios: A Pilot Study · medRxiv (Preprint), December 2025 — Jagarapu J, Babata K, Chamarthi S, Hoyt R
- OpenEvidence, the 'ChatGPT for doctors,' doubles valuation to $12 billion · CNBC, January 2026
- The Use of an Artificial Intelligence Platform OpenEvidence to Augment Clinical Decision-Making for Primary Care Physicians · PubMed Central, 2025
- Large Language Models in Medical Question Answering: Performance and Limitations · Nature Medicine, 2023
- FDA Guidance: Clinical Decision Support Software · U.S. Food and Drug Administration
- OpenEvidence, the 'ChatGPT for Clinicians,' Has Raised $200M on a $6B Valuation and Recently Moved to Miami · Refresh Miami, October 2025
- OpenEvidence Inc. v. Doximity, Inc. et al — Case No. 1:25-cv-11802 (D. Mass.) · U.S. District Court for the District of Massachusetts, Filed June 2025
- Doximity Accuses AI Health Rival of Misinformation, Harassment · Bloomberg Law, September 2025
- Terms of Use (September 17, 2026) · OpenEvidence
- Study of 270 million OpenEvidence queries · JAMA, October 2026
- HIPAA and BAA announcement · OpenEvidence, April 2025
- Visits announcement · OpenEvidence, August 2025
- USMLE announcement · OpenEvidence, August 2025
- Model family announcement · OpenEvidence, September 2026