Model as a Service

Manage and access platform-hosted models through stable interfaces.

Agentic Friendly

AI model operations split into two different concerns:

  • getting model weights or model references into a platform-managed form
  • exposing those models through a serving path that applications can actually use

For locally managed and platform-hosted models, the central operational service is the Model Installer API. It handles artifact import, MLflow registration, inference registration, and the bridge into the serving layer.

In This Section

On this page