Claude Sonnet 3.7 Rate Limit Issue – Please Fix This!

[RebaiFedi]

Select Topic Area

Bug

Body

I talked a little with Claude Sonnet 3.7, then it told me: 'Sorry, you have exhausted this model's rate limit. Please wait a moment before trying again, or switch to a different model.

[github-actions[bot]]

💬 Your Product Feedback Has Been Submitted 🎉

Thank you for taking the time to share your insights with us! Your feedback is invaluable as we build a better GitHub experience for all our users.

Here's what you can expect moving forward ⏩

• Your input will be carefully reviewed and cataloged by members of our product teams. 
  • Due to the high volume of submissions, we may not always be able to provide individual responses.
  • Rest assured, your feedback will help chart our course for product improvements.
• Other users may engage with your post, sharing their own perspectives or experiences. 
• GitHub staff may reach out for further clarification or insight. 
  • We may 'Answer' your discussion if there is a current solution, workaround, or roadmap/changelog post related to the feedback.

Where to look to see what's shipping 👀

• Read the Changelog for real-time updates on the latest GitHub features, enhancements, and calls for feedback.
• Explore our Product Roadmap, which details upcoming major releases and initiatives.

What you can do in the meantime 💻

• Upvote and comment on other user feedback Discussions that resonate with you.
• Add more information at any point! Useful details include: use cases, relevant labels, desired outcomes, and any accompanying screenshots.

As a member of the GitHub community, your participation is essential. While we can't promise that every suggestion will be implemented, we want to emphasize that your feedback is instrumental in guiding our decisions and priorities.

Thank you once again for your contribution to making GitHub even better! We're grateful for your ongoing support and collaboration in shaping the future of our platform. ⭐

[Lokesh-Shrestha]

Sorry, you have been rate-limited. Please wait a moment before trying again.
Server Error: rate limit exceeded Error Code: rate_limited

what does this mean?

[pushpender-singh-ap]

+1 now I am getting the "400 bad error"

[HannibalBarca0]

it's been 4 days, with this rate limit, impossible to use it, paying for nothing. Been quite a long time, with this problem. but no answer, no fix no nothing. Seems better get somewher else with actual support and a working product. Rate limit exceeded on Claude = ALL other models get rate limited

[SuperGamer001]

I'm getting rate limits too. I got school benefits to get Pro, and soon I'm going to need it for my classes soon (AI is part of the class, it's not cheating, lol 😂)

I've been able to get many many prompts out of Claude and o4-mini. Today, I was using o4 for a game I'm making, and I was hit with this. Not only does the o4 stop working, so does every other model. I can't even use Claude 4, which I've never used up until I got rate-limited.

[adrazz-cyber]

yep every message i get the rate limit in VS!!

[clong-viimed]

I'm getting this message more and more often now. The service window is getting shorter and shorter. I used to be able to use it all day (8hrs) before getting the rate limit message just 2 weeks ago. Now I'm getting rate limited after just 1/2 hr of work now. I'm on the companies teams copilot Pro license. It's really starting to become a productivity killer rather than improving it since when I continue after the wait time it has to reread the context history and that shortens the time before the next rate limiting.

[Mogg01]

Just adding that this is happening to me as well, and it's extremely problematic. I was going to upgrade to Pro, but if I'm just going to get rate limited all the time, what's the point?

[Rockhelljumper]

Jan 2026, and this is still happening across multiple models...

[RanojitKumar]

"Sorry, you have exhausted this model" - It's just killing the time!

[franzli72]

This is really bothering... Is this ever going to be increased?

[SuperGamer001]

... Or removed?

[lacvietanh]

pay copilot instead of cursor because of use claude 3.7 with good price, but always Sorry, you have exhausted this model's rate limit. Please wait a moment before trying again, or switch to a different model

[SuperGamer001]

I have tried Cursor since some of the users here seem to like it, and it has actually been doing better, somehow.

[ghost]

Hey I just read and used cursor amazing tool. Github sucks,

[shushu789]

Same, to be honest, this rate limit is too exaggerated, it is almost impossible to use the proxy mode, and can only do some relatively simple work, but it will soon be limited, including the new session window. If I buy pro and cannot make normal requests, then the meaning of buying it does not exist, because for most of the usage scenarios that require internal code prompts or some inquiries are getting fewer and fewer. I hope to solve this problem as soon as possible.

[agileandy]

My rate was expired after less than 3hrs of coding!

Signed up on a Sunday evening and did a bit over an hours work. Monday morning, start coding and an hour later I'm rate locked. Can no longer use CoPilot less than 24 hours after I paid for A MONTH.

What makes this worse is there is no transparency on how many token we are allowed, and how many we have used. GitHub could literally be taking our money and then giving what we pay for an we'd never know.

What makes it worse is the GitHub support on this terrible, with lots of similar posts having no answer.

[igmarketing]

The worst is that we can't even pay more to get higher limit... the pricing plans are broken! Please guys, add a plan with unlimited uses, or with a much higher rate limit.

Same here.. about 3h of coding and can't work anymore.

[zgaduj]

This is a joke with limits... It is better to use another AI agent. But where can I file a claim for a refund? Paid subscription should not have limits as it is written. Where is the authority for abusive marketing?

[Manishbajpai055]

Rate limit hit too early on Claude 3.7 preview (Pro user) with no reset timing shown

Issue Description:
I’m a Pro user, but after just 54 requests, I received the following error:
“Sorry, you have exhausted this model’s rate limit. Please wait a moment before trying again, or switch to a different model.”

There is:
• No clear communication of what the limit actually is.
• No visible countdown or indication of when the limit will reset.

[brunopnogueira92]

Sorry, but what is the model base?

[Manishbajpai055]

That Base model ( Claude 3.7 sonnet) is non reasoning Model which has no limit in pro plan, the selected one is Reasoning model Version of claude which is caped to around 60 chats / 24 hrs window this is what i have witnessed. Limits are not clearly mentioned by GitHub.

[brunopnogueira92]

so if we always choose the ‘Claude 3.7 Sonnet’ version, we have no ‘limit’ because it's a non reasoning model?
Sorry for the questions, I'm just trying to understand the logic.

[Manishbajpai055]

Now they have removed the thinking model from Agent mode. so your current selected model is non reasoning one from Claude 3.7 which has no limit (not clearly mentioned in pricing by GitHub)

[BryarGh]

Now they have removed the thinking model from Agent mode. so your current selected model is non reasoning one from Claude 3.7 which has no limit (not clearly mentioned in pricing by GitHub)

I'm getting limit on agen mode 3.7 , what is happening?

[akaready]

Same. Just signed up for Pro today and barely got an hour or two's worth of interactions with 3.7. Ok, fine, let me use 4o. But no, as reported... rate limit for one model effects all others. 🤦‍♂️

Tried to load my own Claude API key... that also get rate limit issues immediately. I also lost all conversational context with the models with all the switching, and the undo button doesn't allow me to move to a time before the rate limit hit, which means I'm stuck in this ridiculous limbo where the thing I was working on is completely broken, the chat context is all screwed up, and now no models seem to have the memory of what I spent the last hour trying to get it to do.

What a terrible experience. 😔

[ty1919]

Please wait a moment before trying again, or switch to a diferent model? why

[Prenz1983]

I waited for 4 hours, and just after the limit reset, I got a “limit exceeded” warning after only 10 messages. I’ve been trying to contact them, but there’s no way — I keep getting automated responses from an AI! This feels like a scam, and I’m seriously considering canceling my PRO subscription and asking for a refund!

[Aze1337]

I waited for 4 hours, and just after the limit reset, I got a “limit exceeded” warning after only 10 messages. I’ve been trying to contact them, but there’s no way — I keep getting automated responses from an AI! This feels like a scam, and I’m seriously considering canceling my PRO subscription and asking for a refund!

i know its frustrating, please read my comment below.. i think its best for you to pay a fixed price of 200$ usd a month with o1 pro mode, it will save you usage with github copilot agent , or even api costs. if you decide not to, know that 10$ usd a month at moment is a steal, the same usage of the agent with api token directly from claude console would cost LOT MORE than 10$. lol

[Aze1337]

It's strange, it said 7 file edit per 10 minutes as mentioned here : https://code.visualstudio.com/docs/copilot/copilot-edits#_limitations
< Copilot Edits is limited to 7 editing requests per 10 minutes. >
I'm waiting for 10 minutes, I will see what's up and edit this message.

[Aze1337]

it would of course be interesting to see other pricing options, which will def come due to how expensive agents are. we're really fortunate to have it for 10$ USD a month, it's good complementary from cline / sonnet 3.7.

what I recommend is using it to extract files related to the feature / bug you need. (make it write a script that does it for what you need)
then give the output that contain all the files + their content into o1 pro mode ($200 usd membership 100% worth it, its saving so much time and money than directly using sonnet 3.7 with api )

and then you give that to github copilot sonnet 3.7 agent, or cline/sonnet 3.7 to implement exactly what you. need.

writing documentations may sound to be a time waste but actually isn't, it will help a lot.

[Aze1337]

EDIT : the mentioned rate limit on the page are not true.

[Aze1337]

after 1 hour i could use it for 5 messages.

[akaready]

Sorry, @Aze1337 — are you suggesting to use o1-pro at $200/mo instead of VSC + Copilot? Or do you get access to use o1-pro via API without any rate-limiting, and then thus can use it as a model for the agent features in VSC + Copilot? Not up to speed on Open AI's offerings, tbh. 🙏

[Aze1337]

Hello, o1 pro mode in the normal chat gpt version, not the api.
Pro mode in chat gpt is really good imo because o1 pro mode can really think very deeply, better than any model from my experience.
Huge/complex issues that I cant solve with Claude ai sonnet 3.7 or cline/sonnet 3.7 always ends up being solved by o1 pro mode.

So that static fixed cost of 200$ usd a month makes very much sense to me, especially the deep research part. + o1 pro mode unlimited queries.

OPENAI model in cline are terrible, dont bother using them.
OPENAI model in GitHub copilot are present to ask questions, but the agentic part is only 3.5 and 3.7 which is good.
The GitHub copilot that’s 10$ USD a month, I think thats really worth it to save dollars on cline sonnet 3.7!

o1 pro mode with 200$ usd fixed price will give you UNLIMITED queries, so you can plan everything, which saves you lot of queries for github copilot extension. o1 pro mode can save you so many queries , so you wont get rate limiting as much with github copilot, or your own anthropic api credit

[agileandy]

C'mon dude... You cannot seriously be trying to justify a $200 / month cost for most people!

[franzli72]

I used to prefer to split long monolithic source files into modules, but if the limit is on the number of file edits, then it could be worth keeping it in a big file (disgusting solution).

[alvinjamur]

this is complete BS. Microsoft has not changed a bit since the early days - sub par, POS software. the rate limit is another dagger. asking people for $$ and then have software not function. moving to a different working platform. this is utter garbage,,,.,,

[zagangsum]

Are you going to fix this error? Why am I paying for Pro if I'm just going to be rate limited anyway?

[alinangnaiba]

Great! after the trial ends and two days after paying the subscription I now get this rate limit for Sonnet. I thought this is unlimitted?

[Cuconuco]

unlimitted doesn't exists.. even with the max plan you still have a limit

[alinangnaiba]

unlimitted doesn't exists.. even with the max plan you still have a limit

IIRC it was unlimited agent mode when I read it last month. It's the main reason why I signed up. I went back to the pricing page and it's unlimited for GPT4.1. Not sure if I only misread it and did not notice "for GPT4.1" in April or they changed it.

[chriskzn]

The rate limit issue only surfaced once Claude Sonnet 4 was released, prior to that it was a lovely experience.

[TheEnd98]

I get this message every time I use the advanced models:
"Sorry, you have been rate-limited. Please wait a moment before trying again.
Server Error: rate limit exceeded Error Code: rate_limited"
Why ? I use Pro version.

[nkitan]

Same issue for me. I got rate limited on the free version of GH copilot and decided to buy pro. Tried to use claude 4 and i got rate limited again even after buying the pro subscription.

[xinghusp]

Same here.By capturing the packet, I found that the real error was actually 429-Too Many Requests. It seems that Github's server resources are insufficient, but they have no intention of dealing with it. Pro members also encounter this problem

[ashwini-gupta1807]

Sorry, you have exhausted this model's rate limit. Please wait a moment before trying again, or switch to a different model.

I am also a subscribed user but this is disappointing there is no limit mentioned in the docs and also the github support is just totally shit , they are not responding to the tickets raised .

[jinhui2024]

It's just too bad that after coming out of Claude 4, the rate issue has been coming up a lot! It's annoying, and you can't unsubscribe from Pro

[chriskzn]

I wrote a ticket to support as I am still on my 30 day trial period, it seems like they want to use an automated response. The agent usage since Claude Sonnet 4 was released is damaged and causing us all frustration. I hope GitHub will really take us serious as this is a real issue. It gets to searching 3 files and then hits a rate limit, I start a new conversation and then it has forgotten everything of the past. I have asked them to come to this forum to check out that it is a wide problem. I hope they listen and actually take a look here. They suggested I go read this article: 'Sorry, your request was rate-limited.' I had the best experience using Claude Sonnet 3.7 for most of the month until end of last week when Claude Sonnet 4 was released. I would love to know why both models are affected.

[Cuconuco]

They don't reply tickets... their costumer service is trash.

[Cuconuco]

It's just too bad that after coming out of Claude 4, the rate issue has been coming up a lot! It's annoying, and you can't unsubscribe from Pro

What do you mean you can't unsubscribe?

[alinangnaiba]

I've used Claude 4 yesterday in ask and agent mode and I've not experience this issue anymore.

[chriskzn]

It definitely seems to have improved a little, however today I reached the limit again. So what does one do once you reach the limit? It feels like I should hand money over to GitHub for a small option, then hand my extra money over to Claude AI and hopefully get an API key with less restrictions? What are your thoughts?

[thelok1s]

openai gives you more tokens for premium models on a free account at this point, and doesnt just straight up blocks you if you hit the limit.
i am getting rate limited like every 15-90 minutes, and unable to use paid copilot for like HOURS at all (this is not how rate limiting should even work)

[chriskzn]

I guess this post is being sent to the back of the logs. They have closed my ticket on this issue and yet I still encounter issues. Have opted to cancel the Copilot Pro subscription through GitHub and will rather try looking for alternative like Cursor until they can fix their rate limiting issues and there time out of 503 errors. A trick for all you people that do remain on the platform, just let the coding agent know that VS Code gave you a timeout error, there is not a problem with your code but could you please pick up from where you left off. This will let the agent continue without loosing where they were at. Support is definitely lacking at GitHub.

[nguyenph88]

I subbed to Copilot+ specifically to try out 3.7. Unfortunately, after 2 days trying to get the model to work, I give up and request a refund.

What I found frustrating was the tone of the support response regarding my refund. They said:
"You'll be happy to know that we have also issued a full refund for your recent Copilot subscription payment."

Like, WTH??? Given that I spent two days troubleshooting a non-functional product, I'm certainly not "happy" about needing a refund. Is it how they respond to a negative and refund request? We, the customers are always wrong when we request a refund and we should be "Happy" if we receive the refund?

[felipeloha]

What is cursors answer to this?
Claude code can cope with the this and also has a functionality to compress the prompts

[josepharuja]

just switch to another model temporarily - then come back to claude - it could be temporary