GitHub Community
What is the specific limit for "Too Many Requests"? #189990
Replies: 129 comments 149 replies
-
|
This usually means you hit a rate limit but GitHub doesn’t publish an exact number for it. The limit depends on things like
So there isn’t a fixed “X requests per minute” value you can rely on. So Why it happens? And How long it lasts? So What you can do?
So it’s not permanent. Just a temporary rate limit. |
Beta Was this translation helpful? Give feedback.
-
|
Thats what happens when you let Github Copilot vibe code Github Copilot. They probably canceled all their subscriptions to their senior devs. 🤣 |
Beta Was this translation helpful? Give feedback.
-
|
also experiencing issues here, whats going on? also, this issue had my agent running for hours without making progress or me noticing, which cost me many many (paid!!!) actions minutes, very frustrating. would like to be compensated for that to be honest. |
Beta Was this translation helpful? Give feedback.
-
|
This is BS. I know I use a lot but i pay my 47.16 a month and I'm in the middle of damn sending work to a robotics company and get this rate limit shit and I cannot finish. If they cared, they would remove the rate limits for paying customers until they get the system fixed. If I cant work till 100 percent without getting rate limited I will just get used to continue or Cline and API through Venice and use all I damn want without get rate limited. Would be the same price or cheaper. Not as convenient though. |
Beta Was this translation helpful? Give feedback.
-
|
I asked "hi" to claude haiku 4.5 to test if it worked with opencode, and then just got |
Beta Was this translation helpful? Give feedback.
-
|
I'm using gpt-4o-mini and I'm getting a rate limit for utility models. Not even heavy usage. I'm a Pro+ user who's only used 42% of my credits for the month (and it's the 24th). And it's not minutes to wait when this happens, more like 2 hours. I'm using the gpt-4o-mini model! Come on guys, fix this! |
Beta Was this translation helpful? Give feedback.
-
|
I am getting the same request I have pro account myself .. the simple answer github is giving is that they want to be "fair" . So if you are using premium requests heavily ...which you are paying for .. they'll hit you with the Too many request error / You have been rate limited error so that all users can use copilot and they can distribute compute evenly among users |
Beta Was this translation helpful? Give feedback.
-
|
It seems that the approach of "oh my opencode" which might simultaneously request copilot's opus/gemini doesn't work? Before this, when using opus alone, the rate limit was never triggered. |
Beta Was this translation helpful? Give feedback.
-
|
Theres are a few dozen posts/replies on reddit today about that, it has to be a bug(because it makes it non usable), waiting for them to fix it. |
Beta Was this translation helpful? Give feedback.
-
Maybe but probably not because codex is also affected. Specifically, one user who posted a screenshot hit a rate limit in his first 2 prompts ever, having used up only 0.8% of his monthly premium requests(codex 5.3). |
Beta Was this translation helpful? Give feedback.
-
|
That might be related to the recent frantic strategy adjustments, such as canceling the advanced model for student subscriptions. Yesterday, the pro trial also couldn't access the advanced model, but it's back today. Perhaps they are still frantically adjusting their strategies. |
Beta Was this translation helpful? Give feedback.
-
|
If this is indeed the case they shouldn't there be updates on status.github.com - it says all services operational so? |
Beta Was this translation helpful? Give feedback.
-
Sometimes GitHub status page doesn't reflect issues immediately. Could be regional or temporary. Might be worth checking again later or seeing if others report the same issue. |
Beta Was this translation helpful? Give feedback.
-
|
Now it's June 24th and I had this problem both yesterday and today, when only using gpt-4o-mini |
Beta Was this translation helpful? Give feedback.
-
|
Hi everyone, As usage continues to grow on Copilot — particularly with our latest models — we've made deliberate adjustments to our rate limiting to protect platform stability and ensure a reliable experience for all users. As part of this work, we corrected an issue where rate limits were not being consistently enforced across all models. You may notice increased rate limiting, but we are trying to ensure any adjust rate-limits are not impacting a majority of our users, and we expect things to stabilize over the next 24–48 hours. Our goal is always that Copilot remains a great experience and you are not disrupted in your work. If you encounter a rate limit, we recommend switching to a different model, using Auto mode, or exploring a plan upgrade for higher limits. We're also working on improvements that will give you better visibility into your usage so you're never caught off guard. We appreciate your patience as we roll out these changes. |
Beta Was this translation helpful? Give feedback.
-
|
I only use 3% of the req, sitll 97% remaining here, reach the rate limit? |
Beta Was this translation helpful? Give feedback.
-
|
Seems like it's a global problem. It happened to me too, and I've only sent like 3 prompts in the last 5 hours. |
Beta Was this translation helpful? Give feedback.
-
Looks like the 97% is total quota, but the rate limit is per time window (like per hour). So you can still hit the limit even if you barely used your total quota. |
Beta Was this translation helpful? Give feedback.
-
|
is it supposed to stabilize yet? |
Beta Was this translation helpful? Give feedback.
-
|
Now it's June 24th and it's happening again with light usage. |
Beta Was this translation helpful? Give feedback.
-
|
The service has gone from excellent to barely usable in 2 days. The copilot SDK allows you to orchestrate an agentic AI development environment, but the extreme brittleness and lack of clarity of these changes has made it barely usable - If there was more transparency in the limits and I could manage and surface them in my dashboard, then at least I'd be able to tell what's going on. We have a github enterprise subscription, and the experience is decidedly not enterprise right now. |
Beta Was this translation helpful? Give feedback.
-
|
Tbh I would heavily avoid using the sdk until they improve their rate limiting and clarify what their actual policies are for automation and heavy use. |
Beta Was this translation helpful? Give feedback.
-
|
I was considering switching to the higher tier subscription you guys have and was excited to see Microsoft (GitHub) finally doing something right, as I do want to see them be good and successful. This feels like a bait and switch though. I got rate limited and now can't use ANY model. I can stomach losing the money for the subscription, but it sucks to see Github doing what Microsoft does best Microsofting up a good product. |
Beta Was this translation helpful? Give feedback.
-
|
I'm a Pro+ subscriber at 48% usage (on the 24th of the month) and I've been rate limited the last 2 days. Oh well, I guess we'll all need to find a different solution. |
Beta Was this translation helpful? Give feedback.
-
|
This is the kind of thing that pushes users more towards competitors. I have been contemplating for the past few days if to switch over to Claude Code instead of paying for GitHub copilot, but I decided to stay with copilot, but now this.. If this was communicated better, fine. But no, no mention of our wasted money, no mention of an actual solution. But, of course, "If it keeps happening, upgrade! Pay more!" |
Beta Was this translation helpful? Give feedback.
-
|
Paying more doesn't fix it, ironically. |
Beta Was this translation helpful? Give feedback.
-
|
Aye.. "If it keeps happening, upgrade! Pay more!" - it's vexxing. Scaling up ressources for the paying customers seems to not be an option. We're not paying to find our place in a queue system. |
Beta Was this translation helpful? Give feedback.
-
|
I have Annual Pro + same problem, so upgrading is not the solution. They seem to be pushing us to leave and go for competitor options. |
Beta Was this translation helpful? Give feedback.
-
|
agreed. I'm now seriously considering exactly that. |
Beta Was this translation helpful? Give feedback.
-
|
maybe i think there are someone use trick lot of account |
Beta Was this translation helpful? Give feedback.
-
|
Me and my team are currently evaluating a provider for a company subscription. Obvi rate limiting like this is a no-go. What provider do you guys recommend that allows usage beyond the subscription rates and then charges based on usage if needed? |
Beta Was this translation helpful? Give feedback.
-
|
Im looking for exactly this as well. But I'm at a loss at where to begin researching. A confusion point for me, is that i don't know which companys model will be "the best", next month. it changes all the time. So locking in on Claude or OpenAI, does not seem appealing to me. Neither does paying even more, than i already pay for Github Pro. But then again - if i get my code hosted on Codeberg and another agentic AI provider, i'll be saving the sub cost for Github Pro. |
Beta Was this translation helpful? Give feedback.
-
|
It seems we’ve identified a new opportunity to build a business if GitHub fails to meet user needs. |
Beta Was this translation helpful? Give feedback.
-
|
It seems like this highlights an opportunity for alternative providers to step in where GitHub falls short. |
Beta Was this translation helpful? Give feedback.
-
|
Choosing auto-model on Copilot Pro can't solve this problem! please fix |
Beta Was this translation helpful? Give feedback.
-
|
This is just such a case and point that one can never trust a vendor for AI needs. This is as bad as a couple employees silent quitting. Now to move to Claude... |
Beta Was this translation helpful? Give feedback.
-
|
Yep... getting back to Claude |
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
-
|
@k4yraa I don't think you quite understand the issue here that many are complaining about. This isn't |
Beta Was this translation helpful? Give feedback.
-
|
@k4yraa Which organization are you from? Do you represent the official stance? I think with so many replies, there's no need for you to explain anymore. Can't you see what everyone is complaining about? Are you an AI?If you don't know how to speak, then shut up. |
Beta Was this translation helpful? Give feedback.
-
|
So your solution after paying for a Pro + plan is wait? Like the free tier? I really do not think that you know what you are talking about. |
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
-
|
Here comes another smart person, thank you for your overly repetitive and unhelpful answer. |
Beta Was this translation helpful? Give feedback.
-
Agreed, this is not an |
Beta Was this translation helpful? Give feedback.
-
|
If rate limits are triggered that quickly, it suggests something beyond normal usage caps—possibly a temporary issue or edge-case behavior |
Beta Was this translation helpful? Give feedback.
-
|
Time to lean into Claude and Codeberg. Thanks for everything while it was good. Github. It's clear that it was only good when i didn't have to reach out for support. |
Beta Was this translation helpful? Give feedback.
-
|
Yeah, I get your point. |
Beta Was this translation helpful? Give feedback.
-
|
Also hitting rate limits. I got Pro a week ago. |
Beta Was this translation helpful? Give feedback.
-
yep, this is crazy. It stopped me this morning and hasn't let me back in! Now it's totally unusable, already hours since it limited me on Claude models, i can only still use Gemini 3.1, which doesn't cut it for my needs. They are just soft-banning people's accounts for god knows how long. Very incorrect behavior, im also not doing that much at all, and even, im a paying user for my premium requests! |
Beta Was this translation helpful? Give feedback.
-
|
It's insane that this is still happening and still an issue, even after their claim of it being fixed. Even more insane to me is the lack of responses from them about this issue. If there is an issue, fine, but at least be decent enough to address this properly and fully. Especially since this is still happening. Address the ongoing issue, address the wasted token/money of us paying users, address whether or not we will be compensated or not, address what is being done to fix this and when it might be fixed, address the lack of addressing, something! |
Beta Was this translation helpful? Give feedback.
-
|
Yes, still happening, keep facing this issue after 1 or 2 requests :(
|
Beta Was this translation helpful? Give feedback.
-
|
what is going on one request and i am blocked |
Beta Was this translation helpful? Give feedback.
-
|
Dang it! Microsoft/Copilot team did some new bs, it started interfering with my work again just now! 
The "seconds" is fake; every time i wait that amount of time and retry, it quickly comes back with a new amount of seconds. THIS IS STOPPING ME FROM WORKING! I work with the models I choose, it's my token budget and money. C'mon, when will they stop messing with the users, their paying customers? Do they really have a bandwidth emergency and need to do such impactful budget cuts?? I hope this is quickly retracted. If they want do to this, and i can understand bandwidth concerns/high traffic - they should write a quick, full back-off alghoritm, this type of limit really shouldn't stick to the user.. it must stick less than a minute not to risk interfering with our work. EDIT: They have FULLY BLOCKED me from working! This crappy dialog comes back every time i wait some minutes and press "Try again". This is bullocks. Should I now finally cancel my subscription?? Do i need to go outside again and wait for god knows how long just because one of their silly experiments on live users on how they can save some money on bandwidth/tokens?? For real, guys. |
Beta Was this translation helpful? Give feedback.
-
|
Well, i guess no more Copilot for today.. they really made it blocked, its barely workable! Unacceptable. |
Beta Was this translation helpful? Give feedback.
-
|
I have had success by starting a new session. OK, you loose your context, but you can have it make a plan and continue in a new session. |
Beta Was this translation helpful? Give feedback.
-
|
I have an account and I pay $10/months. If you do not fix it, I am going to cancel my subscription. |
Beta Was this translation helpful? Give feedback.
-
|
ربنا ما يحكمكم على ولايا 😂 |
Beta Was this translation helpful? Give feedback.
-
|
Unfortunately, in the past few hours ive been getting stopped by rate limits more than usual, again, unacceptable behavior from the Copilot team! You're affecting me and stopping me from working (yes you are, because only Claude models suit my work, especially now with the GPT context bug: Error 413). They probably tinkering with the alghoritm again (live) to see what they can get away with. Yes, i had a long-running turn when it started (upwards of 30 minutes) but an agent should be able to invetigate for that long if it's not good enough to quickly find the bug that it by itself introduced in the worktree! The agent decides to investigate that long, it shouldn't affect me and lock me out of service for the rest of who knows how long. First, since yesterday, the agent (Claude only) has been extremly slow, and no, not just peak hours. Then all of a sudden it sped up, during this long invetigation run, and locks me out on limits. Very poor experience overall lately. The issues stacking on each other.. don't they care about quality of service anymore, ran out of chips, or what's going on?? Why present the bill (in terms of inconvenience) of your infrastructure's technical debt to your paying users? |
Beta Was this translation helpful? Give feedback.
-
|
Update: im pretty much blocked. Every time it says "Try again in ... seconds" it's a fake number as usual and even waiting a minute won't avoid the same rate limit from reappearing really quickly, in a matter of seconds in the new turn. So they were like "Nah, you had enough for today, you've been constantly on Copilot" but that's my money, I pay for these requests! Absurd. I don't think i can still rely on Copilot for my needs, there's too many competitors who don't pull stunts like this, and proceed to ignore their paying customers about it. Note that a news article (investigative journalism) came out about these specific practices, and in fact, it quoted many of us from this thread: https://devactivity.com/posts/trends-news-insights/github-copilot-s-rate-limit-debacle-a-blow-to-developer-productivity-and-software-quality/ |
Beta Was this translation helpful? Give feedback.
-
|
@Barrixar Thanks for sharing the link. I have subscribed for Copilot Pro+ roughly 3 weeks ago. If I would have gotten notice of this article I would never have considered an annually subscription of Copilot Pro+. |
Beta Was this translation helpful? Give feedback.
-
|
Same here, using opencode and getting rate limited. |
Beta Was this translation helpful? Give feedback.
-
|
Hi, I’m a beginner exploring GitHub and open source. This discussion was helpful, thanks! |
Beta Was this translation helpful? Give feedback.
-
|
Running into same rate limit issues. Yesterday, I've got frequently this "Oops, you reached the rate limit. Please try again later..." message regardless if I even tried free of charge agents!. But after quitting the Github Copilot for Xcode app and relaunching the app, my requests went through and I was at least to some extend not blocked. But today I'm completely blocked by this message although I still have enough tokens to use. It's so insane how such an unstable system can be delivered to customers without letting them instantly know about the reasons of the breakage. But I guess this crazy ""Oops, you reached the rate limit..." is simply not the source of truth, just a stupid standard context free message |
Beta Was this translation helpful? Give feedback.
-
|
I am paying per premium request, but still getting rate-limited, does GitHub just not want my money? This is infuriating... 
|
Beta Was this translation helpful? Give feedback.
-
|
I was on pro+. Resorted to codex cli. |
Beta Was this translation helpful? Give feedback.
-
|
same here. Im paying premium overusage just to use sonnet / opus whatever and getting rate limited... |
Beta Was this translation helpful? Give feedback.
-
|
i hope it's a bug |
Beta Was this translation helpful? Give feedback.
-
|
Anybody using Claude Code Extension for Visual Studio ? Does it work with VS2026? |
Beta Was this translation helpful? Give feedback.
-
|
I’m also interested. |
Beta Was this translation helpful? Give feedback.
-
So now it's a weekly rate limit that I have to wait for 2 days.
This is also not consistent. I got this already a few times, and each time the wait time was different and incorrect. Once it said that I hit my weekly limit and that I can resume in ~2 hours, I tried past that time - got weekly rate limited again... Happened a few times now. And now I need to wait for 2 days. What is the point of paying for a subscription that you can't even used 100% of? I am on around 70% of my monthly usage, at the rate I am getting rate limited, I won't be able to even use up what I am paying for... |
Beta Was this translation helpful? Give feedback.
-
|
● GitHub Copilot Rate Limit: "Too Many Requests" Specific Limit Values GitHub Copilot does not publicly document exact numeric rate limits. The limits vary by:
On the Free plan, premium model usage (Claude Sonnet, GPT-4o, o1) is capped at a fixed monthly quota (e.g., ~50 premium requests/month as of early 2025). Why It Was Triggered You hit the per-model request quota for the model you were using. In opencode (or any third-party Copilot client), this happens faster because:
How Long Until It Lifts
What You Can Do Now
The exact limit thresholds are intentionally undisclosed by GitHub to prevent gaming, so the only reliable signal is the error itself. |
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
and others
Uh oh!
There was an error while loading. Please reload this page.
-
Select Topic Area
Question
Copilot Feature Area
General
Body
I was using GitHub Copilot in opencode and triggered the following warning: “Too Many Requests: Sorry, you've exhausted this model's rate limit. Please try a different model. Please review our Terms of Service.”
What is the specific value of this limit, why was it triggered, and how long will it take to be lifted?
Beta Was this translation helpful? Give feedback.
All reactions