CLI
Everything the console does, from the terminal.
One binary. Launch a slice of a GPU, start a fine-tune, list models, upload data and deploy endpoints. It signs in with your Podstack account and bills to the same wallet.
Install
One line.
terminalsh
curl -fsSL podstack.ai/install.sh | shIn practice
Four things you will do first.
terminalsh
$ curl -fsSL podstack.ai/install.sh | sh
$ podstack login
✓ signed in as you@company.comReference
Top-level commands.
| Command | What it does |
|---|---|
podstack gpu | Launch a GPU or a share of one and SSH in |
podstack train | Managed LoRA and QLoRA fine-tuning jobs |
podstack models | Browse and call the Inference catalog |
podstack code | AI coding agent with live cloud sandboxes |
podstack files | Upload datasets and artifacts |
podstack deploy | Ship inference endpoints and apps |
FAQ
Questions, answered.
Which platforms does the CLI support?
The CLI supports macOS and Linux via the install script, and Windows through WSL using the same script inside the Linux environment.
How does the CLI authenticate?
Run podstack login once; it stores a token for your Podstack account and every command bills to that account’s wallet.
Can I script it?
Yes, every command has a non-interactive form with flags, and the Python SDK exposes the same operations for notebooks and CI.
Does the CLI show prices?
It shows the rate for a GPU and share in your region before you launch, the same number the console shows; the website does not publish a rate card.