Choosing a model and reasoning depth

Model picker#

Tap the model name in the composer.

• Search models by name; models are grouped by provider, with Favorites at the top (tap the star to add or remove).
• The CX AI card shows your quota; the Offline card links to offline models.
• Badges show abilities (tools, reasoning, image input).
• Limit reached: a provider that hit its limit is marked. "Every provider is at its limit. Add another one to keep going." links to Add provider.

The chat remembers its model. Your global default is in Settings → AI models → Default models.

Reasoning depth#

For models that support it, the reasoning picker sets Off, Auto, Low, Medium, High, XHigh or Max. Not all models support depth adjustment.

• Show Thinking Content and Auto Collapse Thinking (Settings → Chat and display → Messages) control how the "Thought for n seconds" block appears.

Per-chat system prompt#

The Conversation System Prompt card overrides the assistant's system prompt for one chat. Use Clear to go back.

Per-model settings#

In Default models you can also pick a Thinking Budget for some models, and on each provider page set abilities and overrides (see Bring your own API key).