It would be good if Anthropic provided some kind of feedback or even toggle to auto-route requests for models being used at thinking levels that would be a better value using a different model.

Sort of like, getting an automatic upgrade at a car rental or hotel if there is availability.