Skip to content

Server command

Cmd: uni

This command builds and starts the UNI server in the foreground.

Parameter Description
--port <n> Port number (default: 6001)
--models-dir Override model storage directory
--dev Run in development mode with hot reload
--debug Enable debugpy on port 5678

To see all available options, run uni --help.

Options for llama.cpp

Arguments prefixed --llama: are passed through to the llama.cpp server:

uni --llama:main-gpu 1 --llama:mlock

Refer to the llama.cpp server docs for available options.