Choose a model or agent
This guide shows you how to pick what the assistant runs on for a chat: a preset, an agent, or a specific model.
The button next to Add context (+), under the message box, shows your current choice. By default it shows Default.
Pick a preset
-
Under the message box, select the current choice (for example, Default).
-
Select one of the following:
-
Default: fast, cost-efficient, and recommended for everyday work.
-
Thinking: a higher-capability model for advanced, complex reasoning.
-
Your workspace decides which model backs each preset.
Pick an agent
An agent is the assistant with its own instructions, and sometimes its own model and permissions.
-
Under the message box, select the current choice.
-
Select an agent from one of these lists:
-
Provided by IT: agents your administrator assigns to your workspace.
-
My agents: agents you created or imported.
-
A list only appears when it has agents in it. To create an agent, select Build a new agent. For more information, see Create your own agents.
Pick a specific model
-
In the message box, type
/model. Select /model in the list that appears, and press Enter.The 🤖 Select Model dialog lists the models you can use, grouped by provider. -
Select a model. To find one, type in Search models....
The choice applies to this chat. If no models are listed, connect a model provider first. For more information, see Connect to a model provider.
You can also change the model in the chat's settings: at the top of the chat, select More actions (⋯), and then Chat settings. Leave Model as Default model to use the default.
Use a saved prompt with its own agent
A saved prompt can remember which agent to use. For more information, see Save prompts you reuse.