still hit the rate limit for the pro plan

Discussions
1

seriously？Why do I still hit the rate limit for the pro plan since the agent requests are unlimited?

Version: 1.1.3
VSCode Version: 1.96.2
Commit: 979ba33804ac150108481c14e0b5cb970bda3260
Date: 2025-06-15T06:35:49.230Z
Electron: 34.5.1
Chromium: 132.0.6834.210
Node.js: 20.19.0
V8: 13.2.152.41-electron.0
OS: Darwin arm64 24.5.0

limit
limit912×456 37.4 KB

2

Hi @Alec_Chan and welcome to Cursor Forum

The new plan is no limit on how many requests you can use per month, but there is a rate limit (how fast you can use it).

See details:

3

I’m curious how you hit the limit? This was me yesterday:

image
image1107×472 35.7 KB

4

thanks, and I will check it out a few hours later :ok_hand:

1 Like
5

to be honest , I don‘t know…

image
image1888×1212 63.3 KB

6

That definitely looks like a bug @danperks

7

Hello Mr. Terminator

image
image697×410 62.9 KB

Do you know why I have not the request graph available, since I have signed up for Ultra plan?
I loved it so much. :weary_face:

8

lol thats another model :slight_smile:

I saw in some cases the graph not working, will pass the details to Cursor Team

cc @danperks

2 Likes
9

We are already working on solving this issue, thank you.

3 Likes
10

Any true information or details about “when or what situation will hit rate limits” is a good respond btw.

11

I do not have that information as mentioned in other threads where you asked the same.

12

That’s what I am also having. Very annoying dashboard bug!

13

Thanks! I’m having that issue too!

1 Like
14

Thanks as Dean mentioned Cursor Team is looking into that.

15

We are currently seeing some ongoing issues with our database infrastructure, which may be causing issues both in-app and on cursor.com around usage reporting.

Please keep an eye on status.cursor.com as that may be having an effect here!

We are aware of the bad reporting issue for Ultra plans on cursor.com - this is being worked on by the team too, and should be fixed soon.

2 Likes
16

屏幕截图 2025-06-19 202614
屏幕截图 2025-06-19 2026141287×485 30.1 KB

I am a pro plan,but i could’t change my price opt.

17

Hey, please send an email to [email protected].

1 Like
18

If the rate limiting metrics were not a black box to the user, there would be less frustration on our end. Does the governing algorithm details have to be secret ? Can there be a rate limit display showing how many percent of the rate limit you have used in real time ? My codebase employs rate limiting with my api providers I’m interested in how Cursor tracks commpute units

The models page documentation no longer says what the request pricing per model is only the context size. How about puting the request pricing back, and including a rate limit quantifier / factor ? Also auto mode is a no go zone for me. Some models I will never use again, deleting code blocks is a blacklist offense, and it happens frequently - I wont mention names.

If there is new executive members the founders have brought in to the mix to raise revnues, that are not programmers, this could be the real problem here.

Taking a break from coding is fundamental to development - you gotta pull back and think about design objectively, on a regular basis. Less is more.

19

Just out of interest, what models were you using?

20

This is a summary of the information on Cursor-Rate Limits , generated by AI.

Cursor rate limits are based on compute usage, not a fixed number of requests.

Here’s a concise summary of how Cursor’s rate limits work.

TL;DR

  • Cursor no longer uses a hard monthly request quota (like 500 requests/month).
  • Instead, it applies rate limits based on your compute usage, which depends on the model, message length, attached files, and conversation context.
  • There are two types of limits: burst (for short, intense usage) and local (refills every few hours).
  • When you hit a limit, you can switch models, upgrade your plan, or enable usage-based pricing.

Key Points

1. How rate limits work

  • Burst rate limit: Lets you use a lot of compute in a short time, but refills slowly.
  • Local rate limit: Refills fully every few hours (the exact time is not specified).
  • Limits are based on compute, not request count. Using more powerful models or longer messages consumes more compute.

2. What happens if you hit a limit?

  • You’ll get a notification and three options:
    • Switch to a model with higher limits (e.g., Sonnet instead of Opus).
    • Upgrade to a higher plan (e.g., Ultra).
    • Enable usage-based pricing to pay for extra requests.

3. Legacy Pro Plan

  • If you prefer a fixed request quota, you can switch back to the old Pro Plan (500 requests/month) in Dashboard > Settings > Advanced.

4. Transparency issues

  • Many users complain that Cursor does not disclose the exact compute units, refill times, or per-model limits.
  • This lack of transparency makes it hard to plan your workflow, especially for long or bursty coding sessions.

5. Practical advice

  • If you need predictable, uninterrupted usage (e.g., for long coding sessions or AI agent workflows), consider the legacy plan or prepare to use usage-based pricing.
  • Monitor your usage and be ready to switch models or plans if you hit a limit.

Example

If you’re coding for 6 hours straight and using a powerful model, you might hit a rate limit unexpectedly. You’ll have to wait a few hours, pay extra, or switch to a less powerful model.

Source

1 Like