OpenEvidence
Clinical answer engine for verified US clinicians. Cites licensed full text from NEJM, all eleven JAMA journals and 300+ other sources. Free at the point of use, funded by pharma ads.
Overview
OpenEvidence answers clinical questions with citations, and the corpus is the product. Direct licensing agreements cover NEJM, all eleven JAMA specialty journals, NCCN, ACC, AAFP, ACEP and others, alongside more than 300 further publications plus FDA and CDC sources. Access requires verification by NPI number or institutional credentials, with medical students admitted on proof of enrolment. DeepConsult, the research agent released to all verified US clinicians, runs multi-step literature reviews rather than single lookups.
Adoption is the outlier number. More than 757,000 verified US physicians are registered, above 40% of practising doctors, with over 20 million consultations recorded in January 2026 alone across roughly 10,000 hospitals and medical centres. Through 2026 the product moved from a standalone destination into the record itself: Sutter Health in February, Mount Sinai across seven hospitals in March and Cedars-Sinai in May, all inside Epic workflows. A March 2026 addition suggests CPT codes and E/M levels from visit transcripts, which puts it on the edge of Abridge and Nabla territory.
The business model is the part worth arguing about. Access is free because pharmaceutical advertisers will pay well for precise reach into roughly 600,000 American prescribers, and US digital pharma ad spend runs somewhere around $20B to $25B a year. A $250M Series D in January 2026 led by Thrive Capital and DST doubled the valuation to $12B on roughly $100M of revenue, and management has been open about moving toward enterprise per-seat pricing through health system deals. Founded by Daniel Nadler, who previously built Kensho and sold it to S&P Global, which explains the emphasis on exclusive content rather than model architecture.
Key Features
- ✓ Citation-linked answers drawn from directly licensed full text, including NEJM, all eleven JAMA specialty journals, NCCN, ACC, AAFP and ACEP material
- ✓ DeepConsult, a research agent that runs longer multi-source reviews instead of single-turn lookups, released free to verified US clinicians
- ✓ Access gated by NPI number or institutional email, which keeps both the corpus licence and the ad model aligned to prescribers
- ✓ Enterprise deployments embedded in Epic workflows at Sutter Health (February 2026), Mount Sinai across seven hospitals (March 2026) and Cedars-Sinai (May 2026)
- ✓ CPT and E/M coding suggestions generated from visit transcripts, added March 2026
Where it holds
- • Free at the point of use, which is a large part of why 757,000+ verified US physicians signed up, over 40% of practising doctors
- • Licensed full text rather than scraped abstracts, and on clinical questions that distinction shows immediately in answer quality
- • Citations resolve to the source passage, so a clinician can check the claim in seconds instead of trusting the summary
- • Adoption is real and daily, above 20 million consultations in January 2026 across roughly 10,000 hospitals and centres
Where it breaks
- • US-only in practice. Verification needs an NPI number or a US institutional email, so clinicians elsewhere cannot get in at all
- • Funded by pharmaceutical advertising aimed at prescribers, and disclosure alone does not settle the incentive question
- • No public API and no self-host path, so there is nothing to build on and nothing to audit
- • Chart-aware support is still absent. It answers literature questions, it does not read the patient record and surface evidence unprompted
My Take
The lesson for engineers has almost nothing to do with the model. Winning here meant signing NEJM and all eleven JAMA titles, then gating access behind NPI verification so pharma would pay for reach, a licensing and distribution play that no amount of retrieval tuning reproduces. Founder Daniel Nadler ran the same content-first script at Kensho before S&P Global bought it. Answer quality at the point of care is strong and citations hold up under checking, but outside the United States you cannot even log in, and the free tier lasts exactly as long as ads out-earn subscriptions.
Quick Info
- Pricing:
- free
- Openness:
- Proprietary
- Starting at:
- Free to verified US healthcare professionals, including DeepConsult, with no seat cost and no consumer subscription. Revenue comes from pharmaceutical advertising and sponsored research content against a US prescriber base of roughly 600,000. Health system deployments inside Epic are negotiated separately, and the company has signalled a shift toward enterprise per-seat pricing over time, so the free tier is a strategy rather than a commitment.
- Added:
- Aug 2026
- Updated:
- Aug 2026
Use Cases
Judge it on your own work
The notes above say where OpenEvidence holds and where it breaks. The fastest check is your own workload.
Visit website ↗Alternatives to OpenEvidence
Elicit
freemiumLiterature search and systematic review across 138 million papers, with screening and extraction accuracy figures the vendor actually publishes.
Gemini Notebook
freemiumGoogle's grounded research notebook, renamed from NotebookLM in July 2026. Upload sources, get answers with inline citations and run code against them in a per-notebook sandbox.