Thought i was going crazy! Seems i’m getting the same errors. For the past 24hrs whenever i try and use a space it says: “ZeroGPU queue Waiting for a GPU to become available”.
Then i always get a " ZeroGPU queue timeout No GPU was available after 60s. Retry later".
Same issue here, PRO account (opera8), for the past 3 days (Aug 21-23).
Not just one or two Spaces — literally every ZeroGPU Space I try gives the same result:
ZeroGPU queue timeout No GPU was available after 60s. Retry later.
I’ve tested this on many different Spaces, including two very unrelated ones for reference: kulkas2pintu/wan555 and even the official huggingface-projects/gemma-4-31b-it — same identical error on both. Since that second one is an official Hugging Face Space, this really doesn’t look like it’s specific to any individual Space’s code or config.
Confirming what @Vladikdfddd said above:
- Not high load — Spaces show idle GPU panels.
- Not quota related — I still have quota left and still get timeouts.
- Failed/timed-out attempts still consume daily quota, which is a separate bug on top of this.
I also emailed HF support about this a couple of days ago and haven’t gotten any reply yet.
Given how many of us are hitting this at once, across totally unrelated Spaces, this really seems like a scheduler-level issue on the ZeroGPU infrastructure itself rather than anything on our end. Would appreciate if someone from the team (@cbensimon?) could take a look.
Still not resolved but I’m getting a different error now as shown below:
Error [Errno 2] No such file or directory: ‘/etc/host-cgroup’
Anyone experiencing the same thing?
Seems to work fine now, anyone else?
Not really. Krea 2 and Qwen still fail to work properly.
I only had success with simple image generation so far (free account). But I don#t use any Video generation before the issue is solved. Are there no admins or people responsible for this site? Support? You know, the ones where all your money went to.
I have the same problem since then. What happens?
Morning time at India it worked properly but again now (night) same issue however for paid cases bit better as working (a bit slow allocation) Looks like HF is putting more restriction for free cases Sad
Many are pro users (me included) 
Same here! Just got PRO, but I keep hitting queue timeouts on every generation. Even though nothing actually worked, my daily ZeroGPU quota got completely eaten up (46.8/40 mins). Really hope this gets fixed and reset soon!
48/40 means? How it possible Hope they are not charging per min post quota over
I tried to change to CPU-free tier and return to Zero. It worked 
what is happening. i am getting same errors.
Hi,
Yes, this can happen with ZeroGPU, especially during periods of high demand. Allocation is dynamic, so a Space may occasionally fail to get a GPU because available capacity is temporarily exhausted, which can result in reservation timeouts.
A few things you can try:
-
Retry after some time, especially during lower-traffic periods.
-
Make sure the Space is using the correct ZeroGPU configuration and that the GPU-dependent functions are properly decorated/configured.
-
Reduce GPU usage where possible so allocations are released quickly.
-
Check the Hugging Face status page and community discussions to see if there is a wider service issue.
-
If the problem persists consistently, test with a minimal Space to determine whether the issue is with ZeroGPU capacity or your application.
If other Spaces are also failing to obtain allocations at the same time, it is more likely to be a capacity/service-side issue rather than a problem with your Space.
Perhaps similar or another problem/glitch? :
I’ve been experimenting with https://platform.nuovoforte.com, and it might be worth exploring. The platform provides multiple options to access GPU resources across different providers, which could be useful when availability is an issue with ZeroGPU.
Might be worth giving it a try and seeing if it works better for your workload.
I thought I was the only one facing this 
ZeroGPU has been a great feature, but getting an allocation sometimes feels like waiting for a seat in a crowded lab.
From what I’ve noticed, the issue is usually related to high demand rather than a problem with the Space itself. A few things worth trying:
Restart the Space/runtime and try again
Double-check that the Space is correctly configured for ZeroGPU
Test during lower traffic hours when possible
Check whether your account quota or previous sessions are affecting availability
It would be helpful if you share the Space link and the exact message/error you’re seeing. That way we can figure out whether it’s a queue issue, quota limitation, or a configuration problem.
Hopefully Hugging Face can improve the visibility of queue status in the future because knowing why a request is waiting would make troubleshooting much easier.