API Limits are a real bummer

Ideas Feedback
1

The API limiting on Pro plan makes cursor nearly useless if want to use a large Frontier model. I get there are real costs to using Claude and GPT but dang, Composer just doesn’t do as well. I just had my account reset yesterday, barely used, and already at 10%.

I hate to be stuck to a particular model but Claude Code starting to look like a more appealing option even though been happy using only Cursor for the past year.

3

Hi @gitcagey Thank you for the post. I have some recommendations here that may help your experience.

  • Use Plan mode before jumping into edits.
  • Start fresh chats more often. Long conversations balloon context fast, which eats into your usage. If you have a need for the prior chat history, you can stay in the same chat, otherwise it can help to start fresh (both from a performance and a cost standpoint).
  • Use more efficient models for simpler tasks. A common approach here is to use more expensive models for planning and complex tasks, and using smaller models for the actual implementation of those tasks.

If you’d like to use a little more usage, you can enable on-demand usage and set a limit, e.g. $15. Additionally, you can always upgrade to a higher tier plan if you find you’re consistently using more than what the Pro Plan offers.

5

I’m facing same issue, I’ve subscribe it on 30th April but API limit has over in a single prompt although the prompte still not completed, how can I use this now? And what are the other benefits of this 20$ Pro plan and what it difers from Free user?

6

Hi @msaifulcsse Thanks for the post.
Cursor’s $20 plan includes at least $20 of included API usage for named models, including frontier models like Claude, ChatGPT etc, in addition to generous usage on Composer and Auto models included in a separate usage pool.

Cursor’s free plan includes limited agent and tab completions. It’s highly recommended that you have a pro plan at a minimum to take advantage of more usage capabilities.