One project, three ways to use it
The goal is simple: store large files and reuse the parts already stored. Run openxet yourself, embed it behind your own product, or use hosted openxet when it launches. The hosted service will run the same released open-source build.
Today, the storage-only server and client are available from source. The Hub and Git layers and hosted service are planned. The documentation explains what works now.
Independent, with shared foundations
openxet implements the Xet protocol developed by Hugging Face. It is an independent project, not affiliated with or endorsed by Hugging Face. The code is available under the Apache 2.0 license.
People behind the code
The project is maintained by Jannes Stubbemann. Decisions, implementation work and contributions are tracked in the public repository. We want feedback from people running storage, building platforms and trying to move their first large file.
Help shape what comes next
Tell us where a workflow gets difficult, report a reproducible bug, improve an example, or work on the implementation. A useful question is a contribution too.
- Share a question or report a bug →
- Read the contribution guide →
- Read how the project is governed →
- Explore the design system →
For a security concern, use the private security reporting route. Keep credentials and personal information out of public issues.