Hello, I’m an 18-year-old aspiring game developer and I’ve been using Cursor for a long time. I wasn’t aware of the new subscription changes Cursor introduced until today, when—while developing my game—it told me I’d hit the rate limit. It asked me to either change my plan or use the Auto model. When I checked the pricing, it’s about $60 USD, which is an amount I simply can’t afford.
Since I’m still learning, using AI really helps me improve, but with this new pricing it seems I won’t be able to keep using Cursor anymore. Because I’m still in the learning phase, I sometimes write incorrect prompts, and I can’t always express exactly what I want, so I end up using more prompts than I should.
Under the old plan, I would use up 500 requests by the fourth week, and I was slowly continuing to develop my game with the slower requests. I’m now in my third week and I don’t think I’ll be able to continue developing. The Auto models’ performance is very poor and they do more harm than good for me. I wanted to hear the community’s opinion on this situation—do you think this new pricing is affordable for young developers like me?
Hi @Yi2222 there are two main options without upgrading your plan
a) Stay with the new plan, make sure to not include too much context (rules, files, …) as now Agent can get the required info by itself very efficiently. Spend some amount learning daily how to improve and you will get better.
This is working well for many users and allows much more than 500 requests per month. e.g. Im writing over 10 000 lines of code per day with Cursor and have not yet reached the rate limit on my personal Pro plan, so it is possible.
b) If you really use Cursor heavily and a lot but prefer to keep the 500 requests limit, you can switch back to the legacy plan in cursor.com Dashboard > Settings.
Feel free to ask here in forum or tag me with @ if you have issues or questions like today. Many are willing to help and give advice.
Here is a bit more info about rate limits.
Thank you for your advice. I’m currently learning something new, and even though I try to avoid using too many prompts, I still end up using far more than necessary. Even if I watch videos or read documentation, nothing really changes, because learning and understanding isn’t that easy. Even knowledgeable people can make mistakes, and since I’m inexperienced, I end up making even more errors.
There is nothing wrong with learning and making mistakes 
Tou dont need to upgrade your plan if you are a bit more careful about how much context you use. e.g. dont attach files, let Agent choose them when needed.
For example you can save some of your request by using simpler models like Gemini Flash, Deepseek or gpt-4o-mini to research and ask questions about your project or to try simpler prompts. They may not be great at coding but they are fine for asking tech questions.
For development which model do you usually use?
Instead of attaching my files, I make sure to keep them in the same folder; I use the Gemini and Claude models in Cursor, and I use Deepseek and ChatGPT from their own websites.
1 Like