For about a month now, I’ve been experiencing serious issues with “Auto” usage across multiple Cursor accounts, both business and personal.
Here’s a rough example: I checked my usage and it was at around 43%, before slowly increasing percent by percent. After just a few more prompts, it suddenly jumped to 100% without any obvious reason or unusually heavy usage. This happened now on Mulitple Accounts.
I’ve never had this problem before. I’ve been using Cursor for more than half a year, often with very heavy token usage across multiple Pro+ accounts, and the Auto usage has never behaved like this.
Something seems to have changed recently, because the usage is now increasing extremely quickly and unpredictably.
Generally the available usage seemed to have dropped massively.
Please let me know what you think, my cursor experience is dropping in a freefall recently and not only because of this.
For AI issues: which model did you use?
Model name (e.g., Sonnet 4, Tab…)
For AI issues: add Request ID with privacy disabled
Request ID: f9a7046a-279b-47e5-ab48-6e8dc12daba1
For Background Agent issues, also post the ID: bc-…
Additional Information
Add any other context about the problem here.
Does this stop you from using Cursor?
Yes - Cursor is unusable
Sometimes - I can sometimes use Cursor
No - Cursor works, but with this issue
The more details you provide, the easier it is for us to reproduce and fix the issue. Thanks!
I checked your accounts and the usage is being recorded correctly. What you saw is how the two included pools work together. Your plan has a Cursor Models pool (which Auto draws from) and a separate, smaller Other Models pool. Once the Cursor Models pool is fully used, Auto keeps working and counts against your Other Models allowance instead, and because that pool is much smaller it fills up after relatively few requests. That is the jump to 100% you noticed.
On the broader change: since late August, Auto is billed at the price of the model each request is routed to, rather than a flat Auto rate. For long agent sessions this uses the included allowance noticeably faster than before, which matches the timing you described.
A few things that help stretch it: start a new chat for each task (long chats resend a lot of context on every step), avoid Fast variants, and pick Composer 2.5 directly for simpler work.
I had a similar experience, but it seems to me that the use of tokens for the types of operations I use most frequently has increased. I will follow your recommendation to open a new chat for each task but it seems to me that my quota will end up in 10 to 15 days. compared to last month i ended up with 60% usage only.