What began as Maziyar Panahi's solo lunch-break effort in July 2025 has evolved into a community-backed Hugging Face organization hosting roughly 1,500 Apache 2.0 medical NER models aimed at fully local clinical text processing and PII removal. Social posts now trumpet 340 million downloads in the first year plus 650 on-device variants, yet the project's own blog and repository list far lower verified figures at the six-month mark.
🚨 A year ago, OpenMed didn't exist.
Today: 340M model downloads. 1,500+ open medical models, all Apache 2.0. 650+ run on your iPhone, fully on-device.
Started by one person. Now built by a community.
You should own your AI, especially in healthcare. 👇
The standout feature remains MLX-optimized variants that run entirely on Apple Silicon and iPhones without sending any PHI off-device, a direct contrast to proprietary cloud medical APIs.
Primary sources confirm steady uptake and multilingual releases, but the headline 340 million figure lacks matching data from the project's blog or Hugging Face listings, leaving its exact scale unclear.
Users are praising OpenMed's 340M downloads and 1,500 open medical AI models, thanking contributors and crediting Hugging Face plus the open-source community for enabling on-device healthcare progress.
No Digg Deeper questions have been answered for this story yet.
Wow!
In one year…
OpenMed has had 340M model downloads. 1,500+ open medical models, all Apache 2.0. 650+ run on your iPhone, fully on-device.
Started by one person. Now built by a community.
They want to demonize open source platforms.
Open source AI changes the world.
All open, Apache 2.0: http://github.com/maziyarpanahi/openmed http://huggingface.co/OpenMed pip install openmed
All open, Apache 2.0: http://github.com/maziyarpanahi/openmed http://huggingface.co/OpenMed pip install openmed
Huge thanks to @huggingface. Their stack made this possible for one person: Datasets, Inference Providers, and Protected Spaces on upgraded HW for my medical endpoints. They even backed OpenMed with credits.
You MUST build on HF. 🤗
@julien_c @ClementDelangue @mervenoyann
@BrianRoemmele Thank you Brian. This is your thesis in action: own your AI. 1500+ on-device medical models and 650+ of them run fully on your iPhone, offline, the data never leaves the device. Open weights, Apache 2.0. Would love to show you what's coming next.
@MaziyarPanahi Maziyar, thank you for all you do. I will say it, you are changing the world sir.
@BrianRoemmele Thank you Brian. This is your thesis in action: own your AI. 1500+ on-device medical models and 650+ of them run fully on your iPhone, offline, the data never leaves the device. Open weights, Apache 2.0. Would love to show you what's coming next.
@MaziyarPanahi Maziyar, thank you! It would be a high honor sir.
@MaziyarPanahi Congrats Maziyar!
@pelaseyed @superagent_ai hi @pelaseyed thank you! appreciate it! let me have a look. at the moment i use codex security to regularly scan the repo.
@MaziyarPanahi Would love to support you with @superagent_ai
@TheZachMueller Thank you Zach! The love and care in open-source community is unbelievable! It's even better when it comes to AI! 🙌 Let's keep going! 🤗
@BrianRoemmele Thank you Brian! Always appreciated your support for open-source! 🤗 If it's changing anything, it's only because it's open and runs on your own device.
That's great, Superagent focuses on contributors and PRs:
- Security posture in PRs (ci/cd, supply chain etc., other vulnerabilities in the PR itself) - Contributor trust (identifies, pr-sprayers, slop and malign accounts) - Triages incoming security reports through GH advisories - Handles Contributor License Agreements
We protect over 1000 repos on GH, including curl, dotenv, firecrawl, paperclip etc. etc.
You can install it here: https://superagent.sh/app
Feedback appreciated if decide to give it a try!
@BrianRoemmele Healthcare is the first good "sell" to the people, by my estimation. Plus, we have tons of data for fuel.
Let's murder breast cancer. My mom will thank whomever is the victor!
@MaziyarPanahi Incredible results! I hope we can go even further. Can this model be used locally?Thank you for your contributions to the open-source community!
@BrianRoemmele the phone-native stat is the real shock. most medical AI cant even run off a cloud server, let alone locally
@BrianRoemmele Brian, appreciate you sir! 🫡
