Hello Cursor team,
I am currently using the Pro plan, but I keep seeing the message:
“AI Limited when using Pro version.”
I would like to clarify what this limitation means exactly. Why am I still facing restrictions even after upgrading to Pro? Could you please explain the specific limits of the Pro plan (e.g., usage quota, token limits, or feature restrictions) and whether there is any option to extend or upgrade further?
Thank you!
Can you provide a screenshot and more context? I have never seen that and am on Pro too.
Yea, you hit your monthly usage limit.
Click this in the application to see your usage:
or go to your dashboard and check out your usage.
Bonus is extra Cursor has given you beyond your limit.
Try using Auto (free until 9/15) or GPT-5-mini which I think is unofficially free.
i think pro version is unlimited
It is only unlimited if you are using Auto model, but that is not even going to be unlimited after 9/15.
Alas, no.
There are limits on Pro. There is now a much more expensive thing called “Ultra”.
I now only use Cursor for debugging. For making changes I use (Free) Google Jules first, and then Cursor to help resolve any problems. This means that I no longer hit the limit on Pro.
In regards to Jules, how does the free version’s “15 tasks per day” work for you? Is a task a request or series of requests?
Here’s the kickoff prompt for the most recent task I gave Jules. It gives an idea of the complexity of the things it can do (similar to Cursor’s Agent mode.)
It’s worth noting that this repo has both an AGENTS.md and a directory called “devprompts” which the AGENTS.md tells agents to look in. The devprompts directory has a 60-line file about how to create a background service.
I have started, but not finished a service called “FileUploadRequests”.
This needs extending, with both additional SQL tables, DAO method(s), and content in the service and worker.
The intention is to model, but generalise what is done in the “Fundchat” service.
There should be ultimate four tables, and the service and worker should be extended to use them. (Some of the tables will need to be created in setup.sql and a patch file)
file_upload_requests, which stores the file meta data during processing (this sort of works)
file_contents, which stores the BLOB file contents (this sort of works)
file_chunks, which does not yet exist. It should chunk the files in the same way that the fundchat service does
file_chunk_vectors, which should work as the fundchat vectors stuff does.
The add file method and all the tables should also be extended to include an optional userid.
Then create a page in src/frontends/console (with a route) that shows all the files for the currently logged in user, and allows them to upload a new one. Put this on the “Your Apps” menu. There is already a method in the dao called get_for_id that should work, but will need extending as below.
Also create a page on the Admin menu called “Show files” that lists files for all users in the same way. There is already a method on the dao called “get_all” that should work but will need extending as below.
These page should also show whether the file has chunks, and whether those chunks have encodings. (It should use a JOIN against the other tables to work this out.)
These pages should not show the file contents - just the status.
There should be an API that allows a file to be uploaded by POST, and responds with a JSON structure. The API should look for an API key, and generate the userid to pass in from that key. The key should not be passed to the files DAO, which should only use the userid.
There were about 6 followup prompts, being:
Together, they counted as only one of the 15 / day
That’s pretty good for free. I can see the value in using it to avoid any limits with Cursor.
