CA: TBA

Rent a GPU.
Real hardware.
No friction.

Pick a GPU, deploy in seconds. Full SSH access with CUDA pre-installed. Pay per hour, cancel anytime.

Browse GPUs
—cheapest $/hr
—GPUs available
<60sprovision time
99.9%uptime
REWARDFirst-time renters receive 10,000 $COMPUTEGPU tokens — airdropped 24hrs after purchase
Popular hardware

Pick your compute.
Start in seconds.

View all GPUs →
How it works

Three steps.
Nothing more.

01

Pick a GPU

Browse hundreds of available GPUs. Filter by VRAM, price, or location. From GTX 1080 to H100.

02

Deploy instantly

Click “Rent” and your instance spins up in under 60 seconds. PyTorch + CUDA pre-installed.

03

SSH & run

Get SSH access immediately. Full root, run training, inference, rendering. Terminate when done.

Why COMPUTE GPU

Built for people who
just want to compute.

No setup required

CUDA, PyTorch, and common ML frameworks pre-installed. Just SSH in and start working.

Pay per hour

No minimums, no commitments. Terminate anytime. Billed by the hour with transparent pricing.

Bare-metal pricing

Community GPUs at a fraction of AWS/GCP prices. Starting from $0.01/hr.

Instant provisioning

No queues, no waiting lists. Average provision time under 60 seconds from click to SSH access.

compute-gpu — terminal
$ ssh -p 29615 root@213.173.102.12
Welcome to COMPUTE GPU · CUDA 12.4 · PyTorch 2.5 · Ubuntu 22.04
root@instance:~$ nvidia-smi
RTX 4090 | 24GB VRAM | 0% util | $0.34/hr
root@instance:~$ ▌
Get started

Start computing.
Right now.

Pick a GPU and have it running in under a minute. No commitment, no minimum.

Browse GPUs