> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tensorpool.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Clusters Quick Start

> Get your first GPU cluster running in minutes

<Note>
  Clusters are ideal for interactive development, debugging, and building new projects. If you have working code and want to run experiments, [Jobs](/quickstart) are recommended instead.
</Note>

<Note>
  Make sure you've [installed the TensorPool CLI](/installation) and configured your API key.
</Note>

<Steps>
  <Step title="Create Your First GPU Cluster" icon="plus">
    Create a 1xB200

    ```bash theme={null}
    tp cluster create 1xB200
    ```

    For multi-node training, create a 4-node 8xB200 cluster:

    ```bash theme={null}
    tp cluster create 8xB200 -n 4
    ```

    You can also create a cluster with a pre-built container image (includes CUDA, Python, and ML libraries):

    ```bash theme={null}
    tp cluster create 1xB200 --container pytorch
    ```

    <Info>
      See [instance types](/resources/instance-types) for all available GPU configurations and [container images](/features/clusters#container-images) for available images.
    </Info>
  </Step>

  <Step title="Check Your Cluster Status" icon="info">
    The `tp cluster create` command will give you a cluster ID (e.g., `c-abc123`). Use it to check your cluster's status:

    ```bash theme={null}
    tp cluster info <cluster_id>
    ```

    Wait until the status shows `RUNNING`. The output will list your cluster's instances, each with an **instance ID** (e.g., `i-xyz789`)

    <Note>
      If you lose the cluster ID, you can always find it with `tp cluster list`
    </Note>
  </Step>

  <Step title="SSH Into Your Cluster" icon="terminal">
    Once your cluster status is `RUNNING`, grab the instance ID from `tp cluster info` and connect:

    ```bash theme={null}
    tp ssh <instance_id>
    ```

    For multi-node clusters, SSH into the jumphost instance first. From there, you can access worker nodes by name (e.g., `ssh <cluster_id>-0`).
  </Step>

  <Step title="Clean Up" icon="trash">
    When you're done, destroy your cluster:

    ```bash theme={null}
    tp cluster destroy <cluster_id>
    ```
  </Step>
</Steps>

## Next Steps

* Learn about [cluster management](/features/clusters)
* Explore [storage volumes](/features/storage) for persistent data
* Check out [multi-node training](/guides/multi-node-training) for distributed workloads