For Researchers: Access models & data
Reproduce results, run benchmarks, and publish on open, versioned datasets — no gatekeeping.
Medical AI
A free platform to accelerate medical research, enhance patient care, and ensure equitable access to advanced AI tools. Founded to bring the ethos of openness to medicine.
The OpenMedical thesis
Uber was built on a simple bet: treat atoms like bits. Touch glass, a car arrives — the physical world becomes programmable, and software is the instruction set. OpenMedical makes the same bet on medicine: touch glass, care arrives.
No waiting room, no phone tree. Describe your symptoms — or share a lab report or photo — and get careful, plain-language guidance in seconds, day or night.
Getting help is mostly logistics: triage, remedies, finding the right hospital, knowing when to escalate. OpenMedical turns that slow, opaque coordination into a few taps.
A 3D digital twin, mood history, and nutrition data turn your health from guesswork into something you can inspect, measure, and improve — atoms, read like bits.
Reproduce results, run benchmarks, and publish on open, versioned datasets — no gatekeeping.
Fork the models, ship adapters, and fine-tune for your specialty with open weights and tooling.
Bring decision support to the bedside and shape the roadmap with real-world clinical feedback.
Deploy on your own infrastructure, meet compliance needs, and co-develop the standards of care.
Closed models can’t be audited, reproduced, or trusted at the bedside. OpenMedical publishes the whole stack — models, data, tools, and evaluations — so the medical community can verify everything.
Every OpenMedical model ships with weights, training recipes, and evaluation cards — inspectable end to end.
Rigorously de-identified, consent-first datasets with full provenance, ready for research and fine-tuning.
Python & TypeScript SDKs, a unified inference API, and eval harnesses purpose-built for clinical tasks.
Preprints, benchmarks, and reproducible notebooks — so every claim can be checked, not just cited.
A public square where clinicians, researchers, and engineers review changes and steer priorities.
Red-teaming, clinical validation tracks, and transparent incident reporting baked into every release.
Built with the community, evaluated in public, and free to inspect, fine-tune, and deploy.
Flagship clinical language model
A foundation model for clinical reasoning, documentation, and literature synthesis — trained openly and evaluated against physician benchmarks.
Multimodal diagnostic assistant
Imaging-and-text diagnostic support for radiology, dermatology, and pathology workflows — designed to assist clinicians, never replace them.
Medical AI must earn trust the way medicine always has: through evidence, peer review, and accountability. Every OpenMedical release passes clinical validation tracks and adversarial red-teaming before it ships — and every finding is published.
Whether you write code, see patients, run studies, or lead an institution — there’s a place for you in OpenMedical.
What OpenMedical does
Know exactly where, when, and whom to seek care. OpenMedical assesses your symptoms under the Hippocratic oath, flags emergencies first, and points you to the right level of care.
🧭 CARE NAVIGATION
I've had constant knee pain and swelling for two weeks. Do I need to see a doctor?
Swelling and pain lasting more than two weeks can point to an underlying joint issue. I'd recommend an orthopaedic specialist — and if the joint is hot, red, or you have fever, seek urgent care today.
Drop in lab PDFs, prescriptions, X-rays, or photos. OpenMedical reads them with multimodal AI — optical text extraction, reference-range mapping, cross-metric correlation, and trend detection — and explains every number in plain language.
🔬 LAB SYNTHESIS
A 3D anatomical twin of your body that turns vitals into organ-by-organ status. Sync Google Fit and watch heart, lungs, liver, and sleep respond to how you actually live.
Find real hospitals around you on OpenStreetMap — distance, phone, directions, and a one-tap Uber to the door. This demo is live: try it right here.
📍 NEARBY HOSPITALS
A daily mental-health check-in that analyzes your stress across mood, sleep, energy, and connection — with real coping tools and honest guidance on when a human therapist is the right next step.
Your stress trend is easing this week — sleep is doing the heavy lifting. Tonight, try the 4-7-8 breathing wind-down before bed. If the low mood persists past two weeks, let's talk about finding a therapist.
OTC, ayurvedic, or homeopathic — MedSearch finds options for minor issues and tells you the truth about the evidence behind each one, checks your medications for interactions, and lists the red flags that mean 'see a doctor'.
🧪 EVIDENCE BLUEPRINT
Evidence strength shown honestly — homeopathy is always labelled unproven.
MEDSEARCH Find a remedy →OpenMedical is also an API and MCP server. Mint a personal key and give Claude, ChatGPT, or your hospital's own AI real medical tools — ask_doctor, find_remedies, analyze_health_stats. Free for everyone — no subscriptions.
Yes. Every feature — the AI assistant, the BioHack digital twin, MindCare, MedSearch, and EatWell — is free with no subscription. There is a fair-use cap of 50 AI calls per week per account.
No. OpenMedical provides educational health information and always defers to licensed clinicians. In an emergency, call your local emergency number first — the Emergency page lists numbers for 60+ countries.
A 3D anatomical model of your body — male or female, with a pregnancy mode — that turns your vitals into organ-by-organ status you can explore, plus an AI read of what your numbers mean. Try it on the BioHack page.
Yes. OpenMedical ships a public REST API and an MCP server with tools for symptom triage, drug-interaction checks, remedy search, meal analysis, and provider lookup. Setup instructions are on the Developers page.
Every remedy carries an honest evidence label — over-the-counter options cite their evidence base, and homeopathic remedies are always marked as scientifically unproven. Red-flag symptoms that need a real doctor are called out on every result.
Partnership inquiries, dataset contributions, clinical pilots, press — tell us how you'd like to work with OpenMedical and we'll reply personally.