UNI (Universal Pathology Foundation Model)
Save to shortlistLast reviewed Nov 11, 2026 against the sources listed on this page · How we review
At a glance
UNI is a general-purpose, self-supervised vision foundation model for computational pathology developed by the Mahmood Lab, trained on over 100 million images from 100,000+ diagnostic H&E…
- FDA:
- Not on record
- Epic:
- Not on record
- Named users:
- Not on record
- Published study:
- Yes, linked below
Description
UNI is a general-purpose, self-supervised vision foundation model for computational pathology developed by the Mahmood Lab, trained on over 100 million images from 100,000+ diagnostic H&E whole-slide images across 20 tissue types. It serves as a feature extractor for downstream tasks such as ROI classification, slide-level classification, and disease subtyping, achieving state-of-the-art results across 34 clinical benchmarks. UNI is widely used as a research backbone in digital pathology and was followed by the larger UNI 2 model.
Quick answers
Answered only from the sourced records on this page. Questions we can't answer from the record are left out. How we review
- Is there published research on UNI (Universal Pathology Foundation Model)?
- Yes, this listing links a published study. Check whether the patients and setting in the study look like yours.Read the study
Security & contracting
The maker has not reported these details yet. Ask the vendor directly during evaluation.
- Workflow stage
- Not reported
- Integrations
- Not reported
- Data inputs
- Not reported
- Deployment
- Not reported
- Signs a BAA
- Not reported
- SOC 2 report
- Not reported
- HITRUST
- Not reported
- Evidence type
- Not assessed
Readiness details are self-reported by the maker and are not checked by Clinical Intelligence Foundry. Evidence type describes the kind of study or document the listing links to; it is not a rating or endorsement. Confirm compliance documents with the vendor before purchase.
Who's using this
Organizations named in public announcements or by the developer. Inclusion is not an endorsement.
No public deployments found yet (checked Oct 2026). Using this tool? The maker can claim the listing to add customers.
Get weekly updates for Pathology
Up to 5 new items a week: new tools, FDA decisions and studies. Newest first, never ranked. Confirm by email, unsubscribe any time.
Sources
This listing links to primary sources. Follow the links below to check details directly with the developer, regulator or publication.
Website GitHub Repository Research PaperDeveloper
Mahmood Lab (Harvard Medical School / Brigham and Women's Hospital / Mass General Brigham)
Mahmood Lab, Harvard Medical School
Evaluating this tool?
Get a printable packet for your committee: what the public record shows, the gaps to raise with the vendor, and the questions for this kind of tool.
Prepare evaluation packetChecklist for your role: Admin · Physician · IT · Compliance
Compare similar tools →