OpenEvidence
Last reviewed
Jun 7, 2026
Sources
20 citations
Review status
Source-backed
Revision
v1 · 1,744 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 7, 2026
Sources
20 citations
Review status
Source-backed
Revision
v1 · 1,744 words
Add missing citations, update stale details, or suggest a clearer explanation.
OpenEvidence is an artificial intelligence platform for physicians that answers natural-language clinical questions with responses grounded in peer-reviewed medical literature and accompanied by inline citations. Frequently described in the press as "ChatGPT for doctors," it is offered free to verified clinicians in the United States and had grown by early 2026 into one of the most widely used software tools in American medicine. The company was founded by entrepreneur Daniel Nadler, who earlier built and sold the financial analytics firm Kensho. In January 2026 OpenEvidence raised $250 million at a reported valuation of about $12 billion, roughly double its valuation three months earlier. Its rapid adoption has made it a closely watched case study in clinical decision support and in whether generative AI can be deployed responsibly inside clinical workflows.
OpenEvidence functions as a medical information assistant. A physician types a question in plain language, for example a query about drug interactions, a diagnostic workup, or the latest guideline for a condition, and the system returns a synthesized answer with footnoted references to the studies and guidelines behind it. The product sits in the same category as established clinical references such as UpToDate and DynaMed, but it is built on a large language model rather than on manually written summaries, and it is conversational rather than article-based.
The "ChatGPT for doctors" label, used by outlets including CNBC and NBC News, is a shorthand rather than a literal description. OpenEvidence is a separate system from OpenAI's ChatGPT and is not connected to the open internet. Access is restricted to licensed clinicians, who verify their identity using a National Provider Identifier number. The company positions the tool as a reference and information service for professionals, not as a consumer health product or an autonomous diagnostic device.
Daniel Nadler, born in Toronto in 1983, earned a doctorate from Harvard University, where he also studied poetry under the Pulitzer Prize winner Jorie Graham and later published a poetry collection. In 2013 he co-founded Kensho Technologies with the engineer Peter Kruskall while finishing his PhD. Kensho built machine-learning systems that let analysts query financial and economic data in natural language, and S&P Global acquired it in 2018 for a reported $550 million, at the time one of the largest acquisitions of an AI company.
Nadler started OpenEvidence in 2021, co-founding it with the Harvard machine-learning researcher Zachary Ziegler, who serves as chief technology officer. Both founders have said they were motivated by personal experiences with medical error and serious illness in their families. The assistant launched to clinicians in 2023 after the company joined the Mayo Clinic Platform Accelerate program, which gave it access to clinical expertise and data-structuring help. Nadler was named to the TIME100 Health list of influential figures in global health in May 2025. The company is headquartered in Miami, Florida.
OpenEvidence is built to reduce the fabricated answers, or hallucinations, that general-purpose chatbots can produce. Rather than relying on a broad model trained on the open web, the company uses domain-specific models trained on medical literature and retrieves source documents to support each answer, an approach related to retrieval-augmented generation. The platform draws on a corpus the company described in early 2025 as more than 35 million peer-reviewed publications, and every response carries citations so a clinician can check the underlying evidence. The system is designed to decline to answer when the evidence is inconclusive, and it is deliberately not connected to the public internet.
A central part of the strategy has been licensing high-quality content directly from medical publishers. In February 2025 OpenEvidence signed a multi-year agreement with NEJM Group, publisher of the New England Journal of Medicine, covering content from 1990 forward across NEJM, NEJM Evidence, NEJM AI, NEJM Catalyst, and NEJM Journal Watch. In June 2025 it reached a similar multi-year deal with the JAMA Network for JAMA and 11 specialty journals. The company also lists partnerships with the American Medical Association, the National Comprehensive Cancer Network, and other professional bodies, and Mayo Clinic is both an early collaborator and an investor.
The company has promoted benchmark results to make its case. It said its system was the first AI to score a perfect 100 percent on the United States Medical Licensing Examination in 2025, up from about 90 percent in 2023. It added features over time, including a HIPAA-compliant mode in April 2025 and a multi-study research feature called DeepConsult in July 2025. In 2026 it announced a collaboration with Microsoft to bring its content into clinician workflows through Microsoft's Dragon Copilot.
OpenEvidence's growth has been unusually fast for a clinical tool. The company reported that as of July 2025 more than 430,000 verified US physicians had registered, roughly 40 percent of the country's doctors, with about 65,000 new clinicians signing up each month and more than 8.5 million clinical consultations logged monthly. By December 2025 it said registrations had reached about 760,000 physicians and consultations had grown to roughly 18 million per month, climbing toward 20 million by January 2026. The company has stated that more than 40 percent of US physicians use the tool on a typical day and that it is in use across more than 10,000 hospitals and medical centers. These usage figures come from the company and have not been independently audited.
Health systems began adopting it at the enterprise level in early 2026, with Mount Sinai announcing a system-wide integration and Sutter Health pursuing a collaboration tied to the Epic electronic health record. Coverage in outlets such as NBC News framed the tool as something many doctors had quietly adopted, often without formal institutional approval, by logging in directly.
OpenEvidence raised money repeatedly through 2025 and into 2026 as usage grew. After self-funding the early work, Nadler raised a Series A in February 2025 that gave the company unicorn status, then closed three larger rounds in quick succession. The January 2026 round was co-led by Thrive Capital and DST Global, roughly doubling the valuation set the previous October. Reporting placed total capital raised at around $700 million, most of it within a single 12-month stretch.
| Date | Round | Amount raised | Valuation | Lead investor(s) |
|---|---|---|---|---|
| February 2025 | Series A | $75 million | $1 billion | Sequoia Capital |
| July 2025 | Series B | $210 million | $3.5 billion | Google Ventures, Kleiner Perkins |
| October 2025 | Series C | $200 million | $6 billion | Google Ventures |
| January 2026 | Series D | $250 million | $12 billion | Thrive Capital, DST Global |
Beyond the lead firms, the company's backers include Sequoia, Nvidia, Kleiner Perkins, Blackstone, Coatue, Conviction, ICONIQ, Greycroft, Breyer Capital, BOND, Craft Ventures, Meritech, Alkeon, and Mayo Clinic. Reports said the business surpassed $100 million in annual revenue by the time of the January 2026 round, making it one of the faster companies to reach that mark, though several investors and journalists noted that the $12 billion price implies expectations of much larger future revenue.
OpenEvidence is free to clinicians, and its primary revenue source is advertising shown to those clinicians, chiefly from pharmaceutical companies seeking to reach prescribers. Drug makers spend large sums marketing to health professionals each year, an estimated $20 billion annually in the United States as of 2019, and OpenEvidence aims to capture part of that spending by reaching physicians at the moment they are researching clinical questions. The company says advertising is kept separate from clinical answers and that advertisers cannot influence the content of responses. Analysts have also pointed to adjacent opportunities, including continuing medical education credit, which the company began offering in 2025, and the potential value of usage data.
The company's significance lies in how quickly an AI assistant moved from novelty to routine use in medicine, and in the publisher partnerships that gave it a content advantage over rivals. Its rise also drew competition: OpenAI and Anthropic both moved into health applications, and incumbents such as Wolters Kluwer's UpToDate added AI features, making clinical AI one of the more contested arenas in AI in healthcare.
That prominence has sharpened questions about safety and accuracy. Independent researchers have reported that performance can fall on harder, subspecialty cases, with one 2025 preprint finding accuracy well below the company's headline benchmark scores, and even a high USMLE score implies a nonzero error rate. A 2025 survey of physicians found that accuracy and the risk of misinformation were the most commonly cited concerns about clinical AI, ahead of liability and patient trust, and clinicians and researchers have warned about overreliance eroding independent judgment. OpenEvidence presents its tool as a support for clinical decisions rather than a replacement for them, and it has not been cleared by the Food and Drug Administration as a diagnostic device, a status that keeps it in a lightly regulated space even as its influence over real patient care grows.