Forums

I'm working on a project that has an explicit requirement that I need hosting with GPU enabled machine. Meaning heavy-lifting of the app is handled not by CPU but by GPU. Is there such option with PA ?

There's no GPU on PythonAnywhere.

Do you plan to introduce GPU's?

We may in the future if enough people ask for it (I've made a note that you have), but we don't currently have plans to do so soon.

I would love to have GPU enabled machines as well

@mwrigley thanks for letting us know.

I third the request! Would be amazing.

Thanks, noted!

I forth the request. Currently using 2 platforms to get GPU access

I am looking for a GPU option as well to run whisper.

Thanks, noted!

Need gpu badly

Thanks, noted!

Same here. Would be great.

Thanks, upvoted the relevant ticket in our backlog.

Would also love GPU, it's unfortunately totally necessary to run ML inference on large models, and that is going to VERY QUICKLY become a pressing need for a lot of people!

Noted! I've added an upvote from you.

Same here, need GPU

@damiensemler Are you needing a GPU for training or inference??

I'd also like to ask @alvion, @jeroencvlier, @ludwigavonts, @potiemhost1, @BenReilly

It would be great to have access to GPU's for both the training and inference of ML. XGBoost runs fine, but when I try to implement more modern & complex models like TabPFN, it takes 3x as long.

That's interesting to hear. Unfortunately I doubt we'll be able to offer GPU backed webapps / notebooks in the near future

Would it be possible to offer GPU for always-on tasks?

I think @sboyd should have been more general -- we don't currently have plans to offer GPU-backed instances for any kind of code on PythonAnywhere. It's certainly on the list of suggested enhancements, but it's not on our immediate roadmap.

GPU, I need GPU badly!!!

See giles' answer above

I believe in the year of 2024 there are many AI projects and GPUs are required on pythonanywere. It would be a very competitive addition to the current service. I'd strongly suggest to put this on a higher priority. Thank you.

That's all true, bit at the same time it's not easy to provide as shared resource.

we need GPU

Thanks for your feedback.

We need gpu too.

Thanks, noted!

We need a GPU as well!

Noted

We need a GPU please!!

Thanks for the feedback!

A GPU would be great.

GPU offer would be usefull

Thanks the feedback!

GPU will be great on PA and will use it.

Would you use it for inference or training?

A GPU would be great.

Thanks! What would you use it for? We're trying to work out what a feature like that would look like; it would cost at least US$1/hour, so people probably wouldn't want to have it running all of the time. So we're trying to work out the use cases so that we can identify how you might tell our system that you want one, tell it that you no longer need it, and so on.