Skip to main content

Model Hub Overview

The Model Hub serves as a centralized repository for all models that have been registered and are available and ready for deployment and testing.

On this page, models are grouped under their associated projects.

  • Each project is displayed as a row in the list.
  • Clicking the arrow icon on a project row expands a dropdown list showing all models registered under that specific project.

For each registered model, the following details are displayed:

  • Model Name – The name assigned during registration.
  • Model Type – The underlying model architecture or algorithm.
  • Created By – The user who registered the model.
  • Time of Creation – Timestamp of when the model was registered.
  • Status – The current deployment status (e.g., Staging, Deployed, Registered etc).

Figure 17 – ModelHub homepage

Model Status​

Models in model hub can be in three stages:

  1. Registered – All models that are sent to model hub and arent in active deployment have regsitered as the default status

  2. Deployed – Models that are deployed and can be used have the status deployed

  3. Staging – Models that are pulled down from deployment and moved back to being registered are done via the staging step.

Deployment​

Models with a status of “Registered” must be deployed before they can be consumed by applications.

Deployment Process​

  1. Click View Details next to the model you want to deploy.
  2. A Model Details window opens, showing all available information about the model.
  3. Click the Deploy button.
  4. Select the instance where the model should be deployed.
  5. The platform automatically:
    1. Creates a compute instance.
    2. Sets up the model.
    3. Generates a server endpoint linked to the deployed model.

Once this process is complete, the model becomes available for consumption via its unique API endpoint.

Deployed Model Details​

After deployment, the View Details button provides access to complete deployment information.

Key post-deployment features include:

  • Secret Key – Automatically generated after deployment, used to authenticate API requests made to the model.
  • API Access – The deployed model can be queried for:
    • Model Features – Metadata and input structure details.
    • Predictions – Model inference results.

To simplify integration, the platform provides ready-to-use request examples in:

  • cURL – For quick command-line requests.
  • Python Script – For programmatic interaction within applications.

Figure 18 – Deployed details