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

# GPU instances

Latitude.sh GPU Instances provide high-performance virtual machines powered by NVIDIA GPUs for compute-intensive workloads.

## GPU Specifications

Each GPU instance comes with a dedicated, full NVIDIA GPU (1 x NVIDIA GPU). This provides you with the complete processing power of the GPU for your compute-intensive workloads without sharing resources with other users.

<img src="https://mintcdn.com/latitudesh/8gtBT0wGSHUPVY8Q/images/product/22.Docs25Q4.VMdeploy.png?fit=max&auto=format&n=8gtBT0wGSHUPVY8Q&q=85&s=f6fdfa665e25b02c4e10bcd613a69f3b" alt="GPU instance deployment page" width="2004" height="1942" data-path="images/product/22.Docs25Q4.VMdeploy.png" />

## Deploying a GPU instance

Set up your GPU instance with these steps:

<Steps>
  <Step title="Access the GPU instance creation page">
    [Log in to the dashboard](https://www.latitude.sh/dashboard), select a project, navigate to **Virtual machines** in the sidebar and click **Create instance**.
  </Step>

  <Step title="Choose GPU type">
    Select the GPU type and location based on your workload needs.
  </Step>

  <Step title="Select an operating system">
    Choose the operating system for your instance. GPU instances default to ML-optimized images but you can select from all available distributions and versions.
  </Step>

  <Step title="Set billing and instance details">
    Select the billing method (hourly or monthly). Provide a name for your
    instance.
  </Step>

  <Step title="Deploy the instance">
    Click **Create** to deploy your GPU instance.
  </Step>
</Steps>

<img src="https://mintcdn.com/latitudesh/jTa6J8Hv5F4W4vnJ/images/product/23.Docs25Q4.VMpage.png?fit=max&auto=format&n=jTa6J8Hv5F4W4vnJ&q=85&s=133921408132d22cad204ac47c44caa1" alt="Virtual machine monitoring page" width="2940" height="1450" data-path="images/product/23.Docs25Q4.VMpage.png" />

## Logging into your instance

<Steps>
  <Step title="Connect via SSH">
    Go to the instance details page to find your access credentials. Use the username, password, and public IP provided to connect via SSH.

    The default SSH port is 22. Root access is disabled, so log in with the provided username and use sudo to switch to root if needed.
  </Step>
</Steps>

## Power management

To power cycle your virtual machine, go to the instance's overview page and click on the status dropdown. Choose one of the available options.

<Note title="Heads up">
  Shutting down a virtual machine does not pause billing. To avoid charges, you
  have to delete the instance instead.
</Note>

## Monitoring

Track your virtual machine's performance metrics including CPU usage, memory usage, storage operations, and network traffic from the monitoring section on your VM's overview page. For detailed charts and historical data, see [Virtual Machine Monitoring](/vms/monitoring).

## Deleting a GPU instance

<Warning>
  Deleting a GPU instance permanently erases all data and **cannot be undone**.
</Warning>

1. Click **Actions** > **Delete**.
2. Type the instance's hostname to confirm.
3. Click **Delete**.