# Get started

> Install the Nodus CLI, sign in, run a command on a GPU, follow it, get its results and see exactly what it cost.

Source: https://nodus-platform-site.pages.dev/docs/getting-started/
Build revision: 211ad9f836655b1c3a2668c4693e442471f28614

This page takes you from nothing to a finished GPU job and its bill. You need a terminal and a browser. New accounts start with a **$30 starter grant**, so the first runs need no card.

1. **Install the CLI.**

   **pip**

   Terminal window

   ```sh
   pip install nodus-compute
   ```

   The Python package (Python 3.10 or newer) includes the `nodus` CLI.

   **macOS and Linux**

   Terminal window

   ```sh
   curl -fsSL https://nodus-compute.ai/install | sh
   ```

   **Windows**

   Terminal window

   ```powershell
   irm https://nodus-compute.ai/install.ps1 | iex
   ```

2. **Sign in.**

   Terminal window

   ```sh
   nodus login
   ```

   Your browser opens to confirm the sign-in. A new account gets an org with a `default` project, an API key for this machine, and the $30 starter grant, which expires 30 days after it is granted. On a machine without a browser, run `nodus login --device` and approve the code from any other device.

3. **Run a command on a GPU.**

   Terminal window

   ```console
   $ nodus run --gpu L4 --image nodus/pytorch -- python -c "import torch; print(torch.cuda.get_device_name())"
   job/run-4kq7z created · est. $0.01–0.03 · starts in ~2–4 min (cold) · Ctrl+C to cancel, -d to detach
    ✓ Scheduled      l4-24g-x1-us · $0.52/h (rate frozen)
    ✓ Provisioning   1m48s
    ✓ Pulling image  22s
    ▶ Running
   NVIDIA L4
    ✓ Succeeded in 2m31s · $0.02 · kept warm 60 s · nodus describe job/run-4kq7z
   ```

   `nodus run` uploads the current directory, prints the estimate before anything is charged, streams the logs and exits with your command’s exit code. The rate is fixed when the machine is chosen and holds for the whole run. Add `-d` to return right away and let the job run on its own.

4. **Follow it.** Every run is a Job you can come back to.

   Terminal window

   ```sh
   nodus get jobs -w                 # live list of your jobs
   nodus logs -f job/run-4kq7z       # stream the logs again
   nodus describe job/run-4kq7z      # status, events and the cost so far
   ```

5. **Get the results.** Declare an output path when you run, then copy it back. Downloads are checked against their SHA-256 digest.

   Terminal window

   ```sh
   nodus run --gpu L4 --image nodus/pytorch --output model=/outputs/model -- python train.py
   nodus cp job/<name>:outputs/model ./model
   ```

6. **See what it cost.**

   Terminal window

   ```console
   $ nodus billing
   $ nodus get usage --field-selector object.name=run-4kq7z --group-by segment
   SEGMENT   AMOUNT
   Boot      $0.018778
   Running   $0.003033
   Teardown  $0.002889
   ```

   You pay for the machine from the moment it is created until it is deleted: starting up (`Boot`), your command (`Running`) and shutting down (`Teardown`), all at the rate frozen at launch. Credit is prepaid: a hold is reserved before a machine starts, and only what was used is charged.

Using a coding agent?

Connect Claude Code, Codex or Cursor to Nodus at [/connect](https://nodus-platform-site.pages.dev/connect/), then ask it to run your command. The agent uses the same account, projects and spending limits as the CLI.

## Next steps

* [Concepts](https://nodus-platform-site.pages.dev/docs/concepts/): resources, projects, and how billing is measured.
* [Pricing reference](https://nodus-platform-site.pages.dev/docs/reference/pricing/): list prices for every GPU, CPU shape, storage and egress.
* [Guides](https://nodus-platform-site.pages.dev/docs/guides/): one guide per feature.