In the old system, 500 requests were enough for me to build many projects over an entire month. Now I have to pay $20 every two days. This is absolutely unreasonable.
Hey, thanks for the feedback. To be candid, we’re not planning to bring back the old 500 requests/month plan. Moving to token-based pricing was a deliberate decision, and we won’t be rolling it back. I get that for new users this can feel abrupt, especially when you hit spend quickly.
A few practical things that can help reduce usage:
- Spend limits: you can set a hard monthly limit so there are no surprises
Spend limits | Cursor Docs - Model choice: different models burn tokens very differently. For simple tasks, the faster models are usually enough. MAX/Opus is best saved for when you really need it
Models & Pricing | Cursor Docs - Context management: the fewer irrelevant files in context, the cheaper the request. .cursorignore and targeted @ mentions help a lot.
Related threads where users share how they optimized usage:
If you’re seeing specific requests that are eating an unreasonable number of tokens, send the Request ID and we’ll take a look at what’s going on.