Once your deployment reaches the Running state, you can connect over SSH. The connection details are shown on the Deployments page, including a command you can copy directly.

​
Connection details

Each running deployment shows:
  • Host: the IP address to connect to
  • Port: the SSH port (usually 22, except on Vast.ai)
  • User: the SSH username, which depends on the provider

​
Which user to use

The SSH username is set by the provider’s image, not by DeployGPU, so it differs by provider: Copy the exact command shown on your deployment rather than typing the user from memory. It is the fastest way to avoid a permission denied error.

​
Connecting

Replace <port>, <user>, and <host> with the values from your deployment details.
You can add this to your ~/.ssh/config for convenience:
Then run ssh mygpu.

​
Which key to use

The dashboard attaches one SSH key to each launch, the key you chose (or the default) on the deploy page.
  • Default key: generated for your account at signup. If you kept it, its private key was downloaded once, as a .pem file, in the ten minutes after signup.
  • Your own key: if you uploaded your own public key under SSH Keys, use the matching private key.
QuantaCloud accepts only RSA or Ed25519 keys. If the key selected for a QuantaCloud launch is ECDSA, the deployment fails with an SSH key error rather than falling back to another key, so keep at least one RSA or Ed25519 key on your account if you deploy there.

​
Opening a template’s application instead

PyTorch + Jupyter, ComfyUI, and Open WebUI + Ollama are reached through the Open Application button on the Deployments page, not through a separate URL shown beside the SSH details. Open Application signs you in with your DeployGPU account and opens a private link that only your account can use. SSH still works on these instances for anything the application itself does not cover. The plain GPU VM template (labelled Bare Metal) has no application link. SSH is the only way in. See Templates for what each of the four installs.

​
Transfer files

Use scp or rsync to move files to and from your instance:

​
Port forwarding

Forward remote ports to your local machine for accessing web UIs like TensorBoard:
Then open http://localhost:6006 in your browser. DeployGPU does not firewall instances, so anything you bind to 0.0.0.0 instead of 127.0.0.1 is reachable from the public internet on some providers, not only through this tunnel.

​
VS Code Remote SSH

You can use the Remote - SSH extension to develop directly on your GPU instance:
  1. Install the Remote - SSH extension in VS Code.
  2. Add your instance to ~/.ssh/config (see the example above).
  3. Open the Command Palette (Ctrl+Shift+P), run Remote-SSH: Connect to Host, and select your instance.
  4. VS Code opens a new window connected to your GPU instance, where you can edit files, run terminals, and debug as if it were local.

​
Troubleshooting

The dashboard marks an instance Running as soon as the machine is up, and on some providers SSH takes a little longer to start answering. Wait a minute after Running and try again.
  • Confirm you are using the private key that matches the public key selected for this deployment.
  • Check file permissions: chmod 600 ~/.ssh/your_key.
  • Confirm the SSH user against the table above rather than assuming root or ubuntu.
  • Confirm the host and port from your deployment details.
  • Check that your local firewall or network allows outbound connections on that port.
The default key’s private half is only available for download once, in the ten minutes after signup. If you lost it, generate a new key pair under SSH Keys, or upload your own public key, and select that key the next time you deploy.