Changelog

Learn about new developer features and improvements.

Updated API - The Machine Learning API now supports multiple models per deployment

The Machine Learning API has been updated with new endpoints that allow you to deploy multiple models, referenced as aliases, within a single deployment.

The new system of aliases allows you to seamlessly swap models without modifying deployment configurations or disrupting downstream integrations.

The new endpoints are:

  • POST /v1/ml/deployments/{deploymentId}/models/actions/add - Add deployed models to the specified deployment
  • POST /v1/ml/deployments/{deploymentId}/models/actions/remove - Remove models from the specified deployment
  • POST /v1/ml/deployments/{deploymentId}/aliases - Create an alias
  • GET /v1/ml/deployments/{deploymentId}/aliases - Retrieve a list of aliases
  • GET /v1/ml/deployments/{deploymentId}/aliases/{aliasId} - Retrieve an alias
  • PATCH /v1/ml/deployments/{deploymentId}/aliases/{aliasId} - Update an alias
  • DELETE /v1/ml/deployments/{deploymentId}/aliases/{aliasId} - Delete an alias
  • POST /v1/ml/deployments/{deploymentId}/aliases/{aliasName}/realtime-predictions/actions/run - Generate real-time predictions using an alias

For more information, see the reference documentation for the Machine Learning API.

3 Apr 2025