I’m trying to understand the rate limit reset mechanism for Claude 4 and would appreciate clarification from anyone who has experience with this.
My specific question: When you hit the rate limit for Claude 4, does the restriction get lifted after:
A specific cooldown period (e.g., hourly, daily reset), OR
Only at the start of the next billing cycle?
I’ve searched through the documentation but couldn’t find clear information about the exact reset timing. This is important for planning usage patterns and understanding when access will be restored.
If anyone has encountered this situation or has official information about Claude 4’s rate limit reset policy, I’d be grateful for your insights.
Thanks in advance!
Note: This post is specifically about Claude 4 model rate limits, not other Claude variants.
It’s so confusing i’m 2 days from next months bill but its saying i should upgrade to pro-plus but does that start over the bill period or just upgrades the current one. One would assume it starts a new one.
Having the same experience here. Last used Claude-4-Sonnet yesterday at 2:30 PM, tried again this morning at 11 AM and still hitting the rate limit. That’s about 20 hours with no reset.
My usage patterns might be part of the issue though - I’m seeing token counts ranging from 11k up to 1.3M per request, with several over the 1M mark. All showing as “Included in Pro” but wondering if the high token volume is what’s keeping me locked out longer.
Getting frustrated not knowing when I can use Sonnet again or what exactly triggers these limits. Makes it really hard to plan development work when you don’t know if your preferred model will be available.
I’m also running into this confusing rate limit situation with Claude 4 Sonnet. My experience:
I received the “You’ve hit your rate limit on this model. You’ve saved $XXX on API model usage this month with Pro. Switch to Auto for unlimited requests or set a Spend Limit to continue with Sonnet.” message.
I waited more than 15 hours after hitting the limit, but still couldn’t access the model, so the cooldown/reset isn’t just a couple of hours in my case.
My usage has actually been pretty light since my last billing cycle reset, so it doesn’t seem to be a monthly quota issue.
The messaging makes it sound like I’m locked out for the rest of the month, but from what I understand (and what’s in the docs), it’s supposed to be a rolling window/cooldown, not a hard monthly cap.
I’ve checked the docs and forum, but haven’t found any official answer about the exact reset period or how the limits are calculated. Some users mention daily or even per-request/token-based limits, but it’s still not clear.
It’s really difficult to plan work when you don’t know if/when you’ll get access back to premium models. Can anyone from Cursor or Anthropic clarify:
Is this a cooldown that should reset after a certain number of hours, or is it tied to the billing cycle/monthly quota?
Are there any tools or dashboards to check your current quota or next reset time?
Has anyone had their access restored after waiting, and if so, how long did it take?
Would appreciate any official clarification or tips from users who’ve figured out a workaround.
Thanks!
Using rate limit is a terrible decision. It makes the product experience terrible. I have to switch back and forth to find a model that can be called. In most cases, even if it can be called, the speed is extremely slow. It is a very bad version.
I’m experiencing the same confusion and frustration with the Sonnet (Claude 4) rate limits.
I purchased an annual subscription at the end of May and have only been using Cursor for about a month. I was under the impression that with the new “unlimited requests” plan, I could use Sonnet more freely. However, after just a few days of normal usage, I received the same rate limit message:
“You’ve hit your rate limit on this model. You’ve saved $XXX on API model usage this month with Pro. Switch to Auto for unlimited requests or set a Spend Limit to continue with Sonnet.”
I don’t consider myself a heavy user, and under the old plan (500 requests/month), I never hit the limit. Now, with the new system, it’s unclear what the actual limits are, how the reset/cooldown works, or when I can use Sonnet again. The lack of transparency makes it really hard to plan my work.
I’ve also checked the docs and forums but couldn’t find any official explanation about the reset logic or how to monitor my quota.
If anyone from Cursor or Anthropic can clarify:
Is this a rolling cooldown, a daily limit, or still tied to the monthly billing cycle?
Is there any way to check when my access will be restored?
Can I opt back into the old 500 requests/month plan?
By the way, my account is currently blocked for web login (though Cursor itself still works), and I haven’t received any response from support after multiple emails.
Would really appreciate any clarification or advice from the team or other users!
It’s useless now they massively nerfed it haven’t been able to use it really at all, just rate limited, looking for alternatives, auto is a cheap model and fecks up code all the time. I turned on usage pricing and got this 31 token-based usage calls to claude-4-opus-thinking, totalling: $60.02 1 $60.02 $60.02!! 60 fecking bucks this is not sustainable using these frontier models. 31 requests cost that much???
I just did this and it shows 15 out of 500 requests used for the month instead of the message that blocks you from getting work done.
Also when you go back there (after reverting) you can opt back into the normal plan if you want.
I suspect reverting to the legacy plan may not be available as an option forever especially as more people do it. So if it was me (which it was about an hour ago when I did this) I’d switch back to the old system if the new one is too difficult for you to work productively in.
