On Subscription & Model Usage for Cost Efficiency

[luyao]

I really like Cursor, but I have two pieces of feedback.

First, regarding subscriptions: I recently upgraded to 1500 rapid requests, and it looks like this level will automatically renew next month. However, I’d prefer to stay at $20 per month, so I even set a calendar reminder to downgrade before the next billing cycle. I checked the forum discussions and realized I might have misunderstood how it works, but I’m still unclear on its mechanics.

Second, about the no-limit model: here’s a screenshot of my usage. You can see that I’ve barely used GPT-4o-mini or Cursor-small because I’m unsure when I can switch to these models without sacrificing content quality. I’d love a product-level solution that takes care of this automatically, so I don’t have to worry about which model to choose to save on costs.

Thanks for the hard work! This is just some simple feedback.

[gambithunt]

I agree on when to switch. I feel like I’m constantly watching the limit of 500.

Right now I have vscode code open with the new ability to use Sonnet 1.5, I keep bouncing between the two. Copilot is unlimited while Cursor for my package is 500. I’d love to just not worry about this.

Still not sure it’s actually a valid worry, I need to hit the limit to see what the slow version entails.