OpenEvidence pricing, free plan and limits

Last updated: 2026-11-02

OpenEvidence is an AI medical search engine for verified US clinicians that answers from licensed journals such as NEJM and attaches a citation to every claim. The company says more than 40% of US physicians use it on an average day, and since September 2026 it offers three models that range from seconds to minutes per answer.

About OpenEvidence

OpenEvidence is a clinical question-answering service for doctors, founded by its chief executive Daniel Nadler and headquartered in Miami. In January 2026 it closed a Series D co-led by Thrive Capital and DST Global at a $12 billion valuation, which the company says brought its total raised to roughly $700 million, with earlier backing from Sequoia, Google Ventures, Nvidia and Kleiner Perkins. The same announcement said the ad-supported business passed $100 million in annual revenue less than a year after the company built a sales team.

Instead of searching the open web, OpenEvidence answers from medical journals and guideline content it licenses, and it signed early content partnerships with the New England Journal of Medicine, the American Medical Association, the cancer guideline body NCCN and the American College of Cardiology. Every answer cites its sources, and since July 2026 an EvidenceGrade label rates the strength of the evidence behind each answer adapting the GRADE approach that Cochrane reviews rely on. In September 2026 the company replaced its single answer engine with a family of models that trade speed for depth: Osler answers in about five seconds and is the default, Sackett takes about half a minute and asks follow-up questions about the patient, and Snow, the successor to DeepConsult, spends several minutes working through the literature before writing a report. A fourth model, Darwin, is available only by application to research partners.

The people who get the most from it are US physicians, residents and advanced practice clinicians who need a cited answer between patients. It also runs inside Epic at health systems including Sutter Health and, since an October 2026 agreement, NYU Langone Health, where answers can draw on the record of the patient currently being seen. Other tools in the release log include Coding Intelligence for diagnosis and billing codes and a hands-free Voice Mode. Our AI research tools guide puts it next to general research assistants, and the medical AI category lists its neighbours.

Access is free for verified US clinicians and paid for by pharmaceutical advertising, so the cost section below is short.

It is not the right tool for everyone. Clinicians outside the United States cannot use it, patients are not the audience, and a doctor who wants help drafting patient letters will find general assistants more flexible: the obvious one being ChatGPT, with Claude handling the same kind of writing, while the ambient scribe Abridge produces the visit note itself. For literature searches that stray beyond medicine, Perplexity casts a wider net. And because the company is funded by drug-company advertising, some physicians will want to check every cited study themselves rather than rely on the summary.

Pricing

Free for verified US healthcare professionals: OpenEvidence describes itself as free to doctors and ad-supported, with no paid plan for individual clinicians. Sign-up asks you to verify your US licence, typically with an NPI (National Provider Identifier) number. All four production models are included at no charge; Darwin is not sold and is offered by application.

Hospitals can also license an embedded Epic version through institutional agreements whose terms are not public. Outside the US there is no route in: trade press reported in April 2026 that access ended in the UK and EU, and openevidence.com showed a location block when Hokai tried to read it from Singapore on 2 November 2026.

Key Features

• Speed-tiered model family: Since 3 September 2026 clinicians choose between Osler (about 5 seconds), Sackett (about 30 seconds) and Snow (several minutes) depending on how much time the question allows.
• Licensed journal sources: Answers draw on content licensed from partners including NEJM, the AMA, NCCN and the American College of Cardiology, with a citation attached to each claim.
• EvidenceGrade: Launched 10 July 2026, it shows a strength-of-evidence label beside each answer so a clinician can tell a large randomised trial from a small observational study.
• Snow deep research: Snow replaced DeepConsult as the long-form option, running parallel literature searches and returning a written report for complex multi-study questions.
• Epic integration: Runs inside the Epic record at systems such as Sutter Health and NYU Langone, where the October 2026 deployment can take the open patient chart into account.
• Coding Intelligence: Suggests diagnosis and billing codes with supporting rationale at the end of a clinical note, aimed at reducing missed reimbursement.
• Voice Mode: Released on 21 May 2026, it lets a clinician ask aloud and hear a cited answer read back, useful when hands are busy.

Pros

• Free access removes the budget conversation that subscription clinical references require, which matters most to residents and small practices.
• Citations plus the EvidenceGrade label let a physician judge the weight of an answer in seconds rather than trusting a bare summary.
• Choosing between a quick model and a deep one fits real clinical time pressure better than a single answer engine did.
• Licensed journal content means answers come from the same sources physicians already trust, not from general web pages.

Cons

• Only verified US clinicians get in; colleagues in the UK, the EU and the rest of the world are blocked, with no announced return date.
• The business runs on pharmaceutical advertising, which some physicians see as a conflict-of-interest risk next to treatment answers.
• Darwin, the model behind the company's strongest benchmark claims, is not available to ordinary users, and those scores are self-reported.
• It is a search and reference tool, not a scribe or patient-messaging assistant, so documentation-heavy work still needs a second product.

Visit OpenEvidence Official Website