I want to use Cursor as a development tool. And just that. I have no desire to switching models every time I switch tasks.
I want the model to be selected automatically based on what I am doing.
If you need more input for automatically chose the correct model give me selectors that indicate task type.
It is fascinating to observer the difference in output of the various models when I play around with it at home, but at work I just want it to fix stuff.
Hey, you can probably get away with just setting one model and leaving it, but we appreciate the notion! We may add an ‘auto’ setting in the future for this.
To add to this—models are constantly evolving, so having an option to use the Cursor team’s recommended model would be great. I’ve seen discussions about which model the internal team prefers, and I’d personally default to that option if it were offered. It’s hard to keep up with the best models, but knowing I’m using what you and the Cursor team use/prefer would eliminate FOMO and uncertainty.
Wanted to suggest this and then saw this existing thread - I think if Cursor had an “Auto” model option, which could pick the best model for the job, and also diverting traffic away from high demand models for tasks that don’t warrant the complexity would be great!
I’ve been despairing about the Claude 3.7 high demand all day; everybody’s going to want to push all their requests through the best model, but surely a smart bit of traffic coordination to better sized models will make Cursor better for everybody!
Other products like Perplexity, OpenRouter have features like this, and it would be easy for Cursor to fold the “Auto” model selection concept into a nice free tier.