To run your code with GPU support, select a base environment with GPU access. If you filter by 'GPU', you'll see the full list (this list is subject to updating over time):
Your code will then automatically execute on one of our GPU machines. This will not incur any special charges to you.
What if my project doesn't fit neatly into one of those bins?
The 'Python with GPU support' image is probably your best bet because it has the conda installer, which is useful for installing languages and frameworks and such.
Ideas for a framework we should support?
Please write to us via live chat or to firstname.lastname@example.org, we'll be happy to hear from you.