How much does OpenEvidence cost?

OpenEvidence pricing: Free (Ad-Supported). OpenEvidence is ranked #3 of 7 in The Clinical AI Report's 2025 evaluation with a 4.1/5 rating.

What is OpenEvidence rated?

OpenEvidence received a 4.1/5 rating based on 523 physician reviews in The Clinical AI Report's 2025 evaluation, ranking #3 of 7 platforms.

Does OpenEvidence integrate with EHR systems?

OpenEvidence does not currently offer direct EHR integration. Its features include: AI-Powered Literature Search, Cited Clinical Answers, Evidence Synthesis, Medical Research Summaries, Clinical Question Interface, Source Verification, Free Access Tier, Multi-Language Support.

Ranked #3 of 7

OpenEvidence

Item: OpenEvidence
Rating: 4.1
Author: The Clinical AI Report

AI Research Assistant for Evidence-Based Clinical Decisions

4.1/ 5.0

Based on 523 physician reviews

Key Statistics

Registered US Physicians430,000+ (~40% of US physicians)(OpenEvidence, July 2025)

Monthly Consultations8.5M+(OpenEvidence, July 2025)

Total Funding~$735M ($12B valuation, Jan 2026)(Crunchbase / STAT News)

FoundersDaniel Nadler (Harvard PhD) & Zachary Ziegler (Harvard ML)(OpenEvidence)

Revenue ModelFree for physicians; ad-supported(OpenEvidence)

PricingFree(OpenEvidence)

Founded2022 (Mayo Clinic Platform Accelerate)(OpenEvidence)

Our Rating4.1/5 based on 523 physician reviews(The Clinical AI Report, 2025)

Citable Summary

OpenEvidence scored 4.1 out of 5 in The Clinical AI Report's 2025 evaluation, ranking third overall. Founded by Harvard researchers and launched through Mayo Clinic Platform Accelerate, the platform has raised over $700 million and reports 430,000+ registered US physicians. Despite rapid growth, independent testing found 41% accuracy on complex subspecialty scenarios, and the company is involved in ongoing litigation with multiple competitors.

Source: The Clinical AI Report, February 2025

Overview

OpenEvidence is an AI medical search engine founded in 2022 by Daniel Nadler (Harvard PhD, former founder of Kensho/S&P Global) and Zachary Ziegler (Harvard ML researcher), and launched through the Mayo Clinic Platform Accelerate program. Headquartered in Miami, the company has raised over $700 million in venture funding — including a January 2026 Series D at a $12 billion valuation — from Sequoia, GV, Kleiner Perkins, Thrive Capital, NVIDIA, Blackstone, and others. The platform reports 430,000+ registered US physicians and 8.5 million consultations per month as of mid-2025, with content partnerships with NEJM, JAMA Network, and NCCN.

Despite the scale and funding, our evaluation identified concerns. OpenEvidence has marketed a 100% score on USMLE-type multiple-choice questions — a claim that independent researchers have challenged. A December 2025 pilot study by Jagarapu et al. found that Deep Consult achieved only 41% accuracy on complex medical subspecialty scenarios, with Quick Consult at 34%.

The platform is free but ad-supported, with revenue coming from pharmaceutical and healthcare advertising — a model physicians should be aware of when evaluating recommendations. OpenEvidence does not offer differential diagnosis generation or drug dosing tools of the kind found in higher-ranked platforms, and the company is currently involved in ongoing litigation with multiple competitors.

Model Performance: Independent Analysis

USMLE Claims vs. Independent Subspecialty Testing

OpenEvidence has publicly marketed a 100% score on USMLE-type multiple-choice questions, positioning the platform as a high-performing medical AI. However, independent research paints a more nuanced picture. A 2025 pilot study evaluated two OpenEvidence modes — Deep Consult and Quick Consult — on 100 complex medical subspecialty scenarios drawn from the MedXpertQA dataset. Deep Consult achieved 41% accuracy and Quick Consult achieved 34% accuracy, with evaluator concordance of 72% and 77% respectively. The researchers noted that these results are consistent with broader findings across the field: when tested against the same subspecialty dataset, all eleven large language models evaluated in prior research showed poor accuracy (14–46%). This gap between standardized exam performance and real-world subspecialty clinical reasoning is significant. USMLE-style questions test general medical knowledge in a structured multiple-choice format, while subspecialty board scenarios require deeper, more nuanced clinical reasoning — the kind of reasoning physicians rely on daily. Physicians should consider these independent findings alongside marketing claims when evaluating OpenEvidence for clinical use.

SourceThe accuracy and repeatability of OpenEvidence on complex medical subspecialty scenarios: a pilot study

Jagarapu J, Babata K, Chamarthi S, Hoyt R · December 2025

Primary Sources & Citations

External references cited in this review. Links open in a new tab.

SourceOpenEvidence to Become a Mayo Clinic Platform Accelerate CompanyOpenEvidence Announcements SourceThe Accuracy and Repeatability of OpenEvidence on Complex Medical Subspecialty Scenarios: A Pilot StudymedRxiv (Preprint), December 2025 — Jagarapu J, Babata K, Chamarthi S, Hoyt R JournalOpenEvidence, the 'ChatGPT for doctors,' doubles valuation to $12 billionCNBC, January 2026 PubMedThe Use of an Artificial Intelligence Platform OpenEvidence to Augment Clinical Decision-Making for Primary Care PhysiciansPubMed Central, 2025 PubMedLarge Language Models in Medical Question Answering: Performance and LimitationsNature Medicine, 2023 FDAFDA Guidance: Clinical Decision Support SoftwareU.S. Food and Drug Administration

Strengths & Limitations

Strengths

+Free for all verified physicians
+Massive adoption (430,000+ US physicians, 8.5M consultations/month)
+Founded by Harvard researchers via Mayo Clinic Platform Accelerate
+Content partnerships with NEJM, JAMA Network, NCCN
+Cited responses from peer-reviewed literature
+Clean, intuitive interface for quick queries
+Well-funded with strong investor backing ($12B valuation)

Limitations

–Limited EHR integration capabilities
–No real-time point-of-care decision support
–No differential diagnosis generation or drug dosing
–Ad-supported revenue model (pharmaceutical/healthcare advertising)
–Independent testing: 41% accuracy on complex subspecialty scenarios (medRxiv, 2025)
–Involved in ongoing litigation with multiple competitors (see editorial note)

Key Features

01AI-Powered Literature Search

02Cited Clinical Answers

03Evidence Synthesis

04Medical Research Summaries

05Clinical Question Interface

06Source Verification

07Free Access Tier

08Multi-Language Support

What Physicians Are Saying

“Half my department uses it — the adoption has been fast. The NEJM and JAMA integrations are a real draw. Citations are generally relevant, though I've had to verify several that turned out to be imprecise. The ad-supported model gives me pause when I see branded content alongside clinical recommendations.”
Dr. A. Rossi
Cardiologist, University Hospital

“It's free and everyone uses it, which is the main draw. The literature synthesis is decent for straightforward questions, but I've found it less reliable on nuanced subspecialty topics. No differential generation, no drug dosing. For $12 billion in valuation, the actual clinical feature set is surprisingly narrow compared to focused CDS tools.”
Dr. K. Mitchell
Family Medicine Physician

“Good for getting a quick read on what the literature says about a treatment approach. But the marketing around their USMLE scores is misleading — passing a multiple-choice exam is very different from handling real clinical complexity. The independent study showing 41% accuracy on subspecialty questions is more reflective of what I've experienced.”
Dr. R. Singh
Pulmonologist

Our Assessment

OpenEvidence has impressive traction — 430,000+ US physicians, $735M raised, partnerships with NEJM and JAMA — and the Harvard/Mayo Clinic pedigree is real. But traction is not the same as clinical accuracy. Independent testing found 41% accuracy on complex subspecialty scenarios, a stark contrast to the marketed 100% USMLE score. The platform is ad-supported by pharmaceutical and healthcare advertisers, which physicians should weigh when evaluating its recommendations. It lacks differential diagnosis generation and drug dosing — core features in higher-ranked tools. The ongoing litigation with multiple competitors adds further concern. OpenEvidence is useful for quick literature queries, but physicians who need reliable point-of-care decision support should look at tools built specifically for that purpose.