The wait is over

Fly GPUs are here

We know you’ve been wanting to use GPUs on Fly.io. We’re happy to announce that they’re available for everyone.

Get started with Fly GPUs

Ready, Set, Go!

Speedrun your app with Fly GPUs

Deploy to a GPU today with just a few changes to your configuration file. Scale up and down in seconds, and only pay for what you use.

Learn more
> Install flyctl on 
Visit https://fly.io/docs/hands-on/install-flyctl/ for installation instructions
 
> Launch an app with attached GPU
fly launch --vm-gpu-kind l40s --from https://github.com/fly-apps/ollama-demo

Not on ? Install flyctl for your platform.

AI/ML workloads at your fingertips

You can recognize speech, segment text, summarize articles, synthesize images, and more at speeds that would make your homelab blush.

NVIDIA L40S 48GB

Choose to accelerate graphics or video workloads, as well as for inference.

$1.25/hr
per GPU
Available regions:
ord
Chicago, Illinois (US)

NVIDIA A100 40GB PCIe

For inference, model training, and intensive high-precision computation tasks like scientific simulations.

$2.50/hr
per GPU
Available regions:
ord
Chicago, Illinois (US)

NVIDIA A100 80GB SXM

For inference, model training, and intensive high-precision computation tasks like scientific simulations.

$3.50/hr
per GPU
Available regions:
ams
Amsterdam, Netherlands
iad
Ashburn, Virginia (US)
mia
Miami, Florida (US)
sjc
San Jose, California (US)
syd
Sydney, Australia
Billed by the second, no minimum usage requirements

We’re only getting started

More coming soon

We are working on getting some lower-cost A10 GPUs in the next few weeks. We’ll update you when they’re ready. If you want to explore the possibilities of GPUs on Fly.io, here’s a few articles that may give you ideas.