I would like to know which model is used in “auto” mode. We get charged differently for different models. Is there a way to restrict usage to certain models for certain tasks. For eg, plan could use any appropriate model, but execution is limited to gemini?
Hey, good question.
Right now, auto mode doesn’t show which exact model it picked for each request. The system automatically picks a model based on how complex the task is, this is called Premium routing.
What you can do:
- Manual model selection: instead of auto, pick a specific model from the dropdown, like Gemini. This gives you full control over what’s used.
- Usage dashboard: on cursor.com/settings you can see which models were used and how much it cost, but only after the fact.
You can’t set different models for the plan and execution phases within the same chat yet. It’s the same model selection for the whole session.
If you need strict cost control, the most reliable option is to pick a model manually instead of auto.