Choose a model
Select an available provider, model, and supported execution controls.
The model picker controls the model used for the requested work. The choices depend on connected providers, workspace policy, and the capabilities the selected model exposes.
Read the visual explanation
An available provider connection supplies model choices. Effort and speed controls depend on the chosen model and workspace policy.
Make a selection
- Open the model picker in the session composer.
- Select an available provider or connection, then a model.
- Set effort, capability, or speed only when the picker offers those controls.
- Review the selection before sending the message.
Higher effort can be useful for a difficult investigation; it is not a guarantee of a better result. Ask for specific checks regardless of the model you choose.
Set up provider access
ChatGPT supplies connected subscription access for Codex sessions. Claude and Cursor have their own connection flows. These connections are separate from Uno MCP, which lets another client operate Uno.
A connected provider can still be unavailable because of account status, provider limits, workspace model restrictions, or a stale connection. Check the message shown by the app and model access settings.
Defaults and follow-ups
Use workspace defaults for the models your team expects for new work. A session or follow-up choice applies to the work being requested; inspect the composer rather than assuming it matches a different session.
If the UI offers a model change on a follow-up, choose it before sending. Do not assume changing a picker replaces an execution already in progress. Keep any important handoff context in the conversation.
For blocked choices, see connection problems. Review usage and limits when choosing a more resource-intensive configuration.