Cloud GPU Machines — No Hardware Required

YOUR GPU RIG
IN THE CLOUD.

Rent a dedicated NVIDIA GPU by the second. Open JupyterLab in your browser or connect over SSH — PyTorch and CUDA are already installed. Stop whenever you're done.

$0.40
Starting Price/hr
Per-second
Billing
100 GB
Disk per Machine
Up to 48 GB
GPU Memory (VRAM)

Choose Your GPU Tier

From 12 GB to 48 GB of VRAM — pick the GPU that fits your workload.

Starter
Notebooks & lighter models
$0.40/hr
  • RTX 3060
  • 12GB VRAM
  • 4 vCPUs
  • 16GB RAM
Deploy Now
Recommended
Standard
24 GB for larger models
$0.60/hr
  • RTX 3090
  • 24GB VRAM
  • 8 vCPUs
  • 24GB RAM
Deploy Now
Pro
Newest GPU generation here
$0.90/hr
  • RTX 4090
  • 24GB VRAM
  • 8 vCPUs
  • 32GB RAM
Temporarily unavailable
Power
Most VRAM — 48 GB
$1.20/hr
  • RTX A6000
  • 48GB VRAM
  • 12 vCPUs
  • 64GB RAM
Deploy Now

How It Works

Three steps to your cloud GPU machine.

01

Choose Your GPU

Pick from RTX 3060 to RTX A6000, 12 to 48 GB of VRAM.

02

Deploy in Minutes

Usually ready in a few minutes, with PyTorch and CUDA installed.

03

Connect & Work

Open JupyterLab in your browser, or connect over SSH.

Built For GPU Work

Compute you run from a notebook or a terminal.

AI / ML Training

Train and fine-tune PyTorch models on 12 to 48 GB of VRAM.

Inference

Run open models with libraries such as Hugging Face Transformers and Diffusers.

Notebooks & Data

Explore data and prototype in JupyterLab, with the GPU one cell away.

Rendering & Batch Jobs

Run GPU jobs from the command line, such as Blender renders in background mode.

GamePC vs Hardware

When renting a GPU beats buying one.

FeatureGamePCOwn Hardware
Upfront CostNone — prepaid credit from $1.15The GPU plus the PC around it
Setup TimeUsually a few minutesBuild it, install drivers and CUDA
Different GPUDeploy on another tierBuy a new card
Hardware UpkeepNone on your sideYours to fix
AccessAny computer with a browser or SSHWhere the machine is
AvailabilityTiers can sell out for a whileAlways yours

Packed With Features

Everything you need, nothing you don't.

Fast Deploy

Usually ready in a few minutes. A deploy that hasn't started within 10 minutes is cancelled and not charged.

Per-Second Billing

Charged for running time only, to the second. Stop whenever you're done.

JupyterLab in Your Browser

One click opens notebooks, a terminal and a file browser on your machine.

SSH Access

Add your public key and run scripts and long jobs from your own terminal.

PyTorch + CUDA Ready

Machines start from the official PyTorch image, CUDA libraries included.

Root Access

You're root in your Linux container — pip or apt-get install what you need.

One Whole GPU

Each machine gets a full NVIDIA GPU to itself, not a slice of one.

100 GB Disk

Every machine gets 100 GB. Kept 24 hours after you stop it, then deleted.

No Contracts

Prepaid credit, no subscription. Unused credit is refundable within 14 days of purchase.

Frequently Asked

GamePC rents you a cloud machine with one dedicated NVIDIA GPU. You open JupyterLab in your browser or connect over SSH, with PyTorch and CUDA already installed, and pay for the time it runs, by the second.

You add prepaid credit, from $1.15. A running machine is charged by the second at its tier's hourly rate, and the charges stop when you stop it. No subscription, no contract.

No. GamePC machines are Linux containers for compute work — AI/ML, notebooks, rendering and batch jobs. They have no display, no graphical desktop and no game streaming, so games can't be played on them.

Linux with PyTorch and the CUDA libraries it uses, JupyterLab and SSH. You are root inside the container, so you can pip or apt-get install anything else you need.

Each machine has a 100 GB disk. When you stop the machine the disk is kept for 24 hours, then deleted; deleting the machine deletes it straight away. Download anything you want to keep.

Yes — deploy a new machine on the other tier. Each machine has its own disk, so copy across any files you need first.

On Every Machine

What you get on every tier.

1 NVIDIA GPU
A whole card, not shared
JupyterLab
Opens in your browser
SSH
Connect from your terminal
PyTorch + CUDA
Preinstalled
100 GB disk
Kept 24 h after you stop
Root access
Install what you need

Ready to Level Up?

Stop waiting for hardware. Deploy a cloud GPU machine — usually ready in a few minutes — and start working today. No credit card required to sign up.