gpu.st Documentation

gpu.st is a GPU cloud platform that lets you spin up GPU instances over SSH. Connect via SSH or the web interface to manage your instances.

Quick Start

Connect to gpu.st via SSH to get started:

ssh gpu.st

If you're a new user, you'll be prompted to verify your email address. Once verified, you can create your first GPU instance:

new

That's it! You now have a GPU instance running. SSH into it:

ssh myinstance@gpu.st

Connecting to Instances

There are two ways to access your instances:

• SSH: ssh instancename@gpu.st
• HTTPS: Your instance exposes port 80 at https://instancename.mygpu.ws

Billing

Instances are billed per hour based on the GPU type. Running instances are charged at the full rate, while stopped instances are charged at a reduced rate for storage. Use the plan command to check your remaining credits.

Commands

When connected to the gpu.st SSH shell, you have access to these commands:

Instance Management

• new - Create a new GPU instance
• ls - List your GPU instances
• ssh - SSH into a GPU instance or run a command
• start - Start a stopped GPU instance
• stop - Stop a running GPU instance
• rm - Delete a GPU instance
• share - Manage instance sharing
• limit - Manage credit usage limits

Account

• plan - Show your current plan and credits
• gpus - List available GPU types
• whoami - Show your user information
• delete-ssh-key - Delete an SSH key from your account
• browser - Generate a magic link to log in to the website

General

• help - Show help information