Edit the model endpoint, generation params, server settings, and compaction budget used by the chat runtime.
Leave as-is to keep the current key or replace it with a new one.
Server settings are saved immediately but only take effect after you restart the app.
Forwarded to LM Studio when the backend supports reasoning controls.
When enabled, live reasoning starts expanded. You can still collapse it while the model is thinking.
This is the estimated prompt size budget used before compaction starts.
Compaction starts above this fraction of the budget.
The summarizer tries to shrink history below this fraction.
Type a message to begin, or select an existing conversation