On-demand GPU instances with SSH access to run any GPU-accelerated workloads and let you run, test, and experiment with any AI application seamlessly.

​
1. Select a base image

Spin up a compute instance by choosing one of the available base images:
  • PyTorch — the latest NVIDIA NGC PyTorch image, preloaded with PyTorch and CUDA libraries.
  • Ubuntu — a vanilla Ubuntu image for a minimal starting point. On full GPU instances, NVIDIA drivers are included via driver passthrough. On MIG instances, NVIDIA drivers are not available yet.
Enter your SSH public key to configure access to the instance, select a GPU instance type, and click Deploy.

​
2. SSH into the instance

Once the instance is ready, navigate to the deployment details page. The Endpoint Configuration section displays:
  • Endpoint URL — the hostname for your instance. Next to it are the copy button (copies the URL) and the SSH button (copies ssh root@<endpoint_url> so you can paste it directly into your terminal).
To connect, use the SSH command with the root user: 
ssh root@<endpoint_url>
The instance comes preloaded with the libraries included in your selected base image. For PyTorch instances, CUDA libraries are bundled in the NGC image. For Ubuntu instances on full GPU hardware, NVIDIA drivers are available; on MIG instances, NVIDIA drivers are not available. Additional packages and libraries can be installed with your preferred package manager.

​
What’s Next

LLM Serving

Explore dedicated public and private endpoints for production model deployments.

Clients

Learn how to interact with the NVIDIA CCluster programmatically

Resources and Pricing

Learn more about the NVIDIA CCluster’s pricing.

Private Inference Endpoints

Learn how to create private inference endpoints

Submit a Support Request

Submit a Support Request.

Agents on CentML

Learn how agents can interact with CentML services.