FOR YOUR OWN HARDWARE
Your stack.
Your machine.
Download the pieces you need. Run a single model server, connect a fleet, or bring your own client.
Explore the apps ↓THE BUILDING BLOCKS
Choose your packages
Debian · published .deb packages.
Start with
SEE THE WORKSPACE
A closer look.
Screenshots from Hugpy, Agent and Station. Open any image for the full view. Interfaces reflect the versions captured and may differ from the current release.
A SMALL START IS ENOUGH
Every machine has a role.
Serve models
Install Wrapper on each inference machine. Add the llama.cpp engine when serving GGUF models.
Connect machines
Install Core on the machine coordinating your fleet. Connect each Wrapper using your Core’s address and registration key.
Choose a client
Harness brings terminal conversations. Station provides the desktop application. Add media packages as you need them.