Feature request for product/service
Cursor IDE
Describe the request
In the models settings, each model only has an on/off toggle. It would be much better if I could also configure each enabled model, for example capping effort at Medium or context at 256K. That way, even when Auto picks a model like Grok 4.7, it would stay within my limits instead of jumping to High, Extra High or Fast variants that burn through a lot of tokens. This would give me real control over cost while still letting Auto choose the model.
