GPUtw Docs

Tutorials

Common GPUtw workflows, step by step.

Launch your first GPU instance

Go to Dashboard → Deploy, choose a GPU, select an available machine, pick a template, review the hourly rate, and deploy. The Instances page updates automatically while the container starts.

Connect over SSH

Add your public SSH key in Dashboard → SSH Keys. After your instance is running, copy the command from the Instances table.

Example
ssh pod-<instance-id>@ssh.gputw.ai -p 2222

Open Jupyter or a Web UI

Use the Web UI button in the Instances table. GPUtw mints a short-lived access URL and then stores a per-instance browser cookie for that UI.

Expose a port

Use the Network/Ports action on an instance. HTTP ports can be private, public, or unlisted with a password. Raw TCP/UDP exposures get an endpoint on tcp.gputw.ai or udp.gputw.ai.

Use an API key

Create a scoped key in Dashboard → API Keys, copy the secret once, and send it as a bearer token.

Example
curl -H "Authorization: Bearer gputw_live_..." https://gputw.ai/api/instances

Stop or delete an instance

Stop keeps the instance record for restart when capacity is available. Delete terminates and removes the instance. Billing stops when the instance is no longer running.