I have used the free cursor version and it is very frustrating to be limited to running a prompt and being told to come back at a so later date as I exceed my limit even though if I go to the usage and see that my current usage was just 60%. I am also new to using Cursor and AI-assisted software design and dev, which is what I want to use it for.
So I decided to look into the Pro plan. Everyone I asked about what plan should I subscribed to suggested that the Pro plan was more than sufficient for what I wanted to use it for as long as I stay with the Composer and/or grok models, and that those models will again be sufficient for software architecture and design conversions, code generation, CI/CD kind of dev work.
However, for the $20 pro plan, it says…
✓Extended limits on Agent
✓Generous limits for Grok & Composer
What does this really mean? Will I run out of request to the Agent and have to wait for weeks, days, a day, a few hours… what?, before I can continue using the Agent?
If I stay within Grok & Composer, how much input/output can I do without incurring additional costs? Does the limit resets daily, weekly, monthly.
As an individual user, I want to set an expectation for myself that my cost is a fixed $20/mo using non frontier models.
Can someone in support (or with similar use case as mine) shed some understanding of what those 2 bullets mean? It will help me decide whether to jump on the bandwagon or rethink my needs. Thank you!
Both bullets are talking about included monthly usage pools. Key points:
The reset is monthly, on your billing cycle date tied to when you started the subscription. It’s not daily or weekly. On Pro there’s no “wait a week” situation, the longest you wait is until the next monthly reset.
Pro has two pools: Cursor models pool (Composer, Cursor Grok 4.5, and Auto) which is what “Generous limits for Grok & Composer” refers to, and it’s much larger. Then there’s the Other models pool (third-party frontier models like GPT, Claude, Gemini) which is smaller. As long as you stick to Composer, Grok, or Auto, you only use the bigger pool.
The $20/month stays fixed unless you explicitly enable on-demand pay-as-you-go usage. It’s off by default. So without you turning it on, nothing over $20 will be charged. If you hit a pool limit, those models will just pause until the monthly reset, they won’t start charging you.
“Extended limits on Agent” is the same Agent as on Free, just with a much higher included limit.
We don’t publish the exact dollar sizes of the pools, but they’re set so typical individual use of Composer, Grok, or Auto usually won’t hit them. For architecture work, code generation, and similar tasks, Pro typically has plenty of headroom.
About your Free experience (60% but still blocked): on Free the pool is very small, and the dashboard sometimes shows a mixed or delayed number while one pool is already exhausted. That’s where the confusion comes from. On Pro the pool is much bigger, so you usually won’t run into this.
Dean, thank you for your response. This is much clearer and helpful. I was wondering what the usage experience is for other independent developers using Cursor Pro plan.