Reference

Settings

Open Settings from the menu bar or the sidebar. Changes save as you make them.

TabHolds
General ConfigurationChat behaviour (title generation, Enter to send), the default model and sampling, the app lock, and your data (export, import, bulk delete).
Remote AccessThe network address and port of Assistant and of the built-in llama.cpp, and the connection history. See Remote access.
EnginesYour inference engines (chat models) and decision engines (the coordinator's Jev) with their models, GPU acceleration, model downloads, the catalogue, the coordinator model, and the Server card: time zone, compaction threshold, knowledge idle minutes, catalogue address, allowed origins. See Models and engines.
PersonasYour personas.
RecipesYour recipes.
Knowledge BaseThe facts Assistant remembers, the memory models and the logs. See Memory.
ToolsThe tools the model may call, with their keys. Time zone and home location are under Basics.
AutomationsYour automations.
About AssistantVersion, update check, changelog, release history, local data paths, keyboard shortcuts, terms of use, report an issue. See Updates and support.

Conversation settings#

The model, sampling, system prompt, tools and tags of one conversation are in the panel on the right of the chat. See Chat.

Setup wizard#

The wizard of the first start asks for the values that have no safe default: the app lock, time zone, home location, the engine and the memory model. Every value can be changed later in the tabs above. See Get started.