AnythingLLM

This is a community package of AnythingLLM, maintained independently. It repackages the official upstream Linux desktop build unmodified.

The application downloads its local inference engines and models into its private data directory. You can also connect a separately running inference server or cloud provider.

Create workspaces, import documents and chat with local or cloud AI models. Connect to inference servers through their network endpoints.

Meeting Assistant transcription and meeting-audio imports are unavailable in this package.

On systems without working GPU rendering, start with: flatpak run com.useanything.AnythingLLMDesktop --disable-gpu

Update the application with Flatpak. Host command execution, host software installation and control of other desktop applications are unavailable inside the sandbox.

Files selected through the file chooser need no general home-directory access. For agent access to your Documents folder, enable: flatpak override --user --filesystem=xdg-documents com.useanything.AnythingLLMDesktop

Undo optional permissions with: flatpak override --user --reset com.useanything.AnythingLLMDesktop