Skip to main content
Auto lets Replit choose a model for the task. To choose one yourself, open the Agent settings dropdown in the prompt box, select Power Mode or Max Mode, turn off Auto, then select Primary model. Free Mode always uses Auto. Builders on Core and Pro can choose an available model in Power Mode or Max Mode. The model list in Replit is the source of truth for your account because availability can vary.
Demo workspace model selector with GPT-5.6 Sol selected in Max Mode and Effort set to Medium

Model and Effort controls in the Agent settings menu.

​
Set Effort for the model

Effort controls how much reasoning the selected model spends on a response. It is a per-model control, not a separate mode. The supported levels run from Low through Max, and each model shows a recommended setting. Use a higher Effort only when the task calls for deeper reasoning. Higher settings can take longer and cost more on complex work.

​
Learn more

Model selector reference

Review Auto, available models, Effort levels, and Compare models.

Open the reference →
Effort details

Compare Low, Medium, High, Extra High, and Max.

Review Effort →
Model and Effort settings shape how Replit responds. Voice Mode changes how you provide the request by letting you speak instead of type.