Baseten provides two sets of API endpoints:

  1. An inference API for calling deployed models
  2. A management API for managing your models and workspace

Many inference and management API endpoints have different routes for the three types of deployments — development, production, and individual published deployments — which are listed separately in the sidebar.

​
Inference API

Each model deployed on Baseten has its own subdomain on api.baseten.co to enable faster routing. This subdomain is used for inference endpoints, which are formatted as follows:

https://model-{model_id}.api.baseten.co/{deployment_type_or_id}/{endpoint}

Where:

  • model_id is the alphanumeric ID of the model, which you can find in your model dashboard.
  • deployment_type_or_id is one of development, production, or a separate alphanumeric ID for a specific published deployment of the model.
  • endpoint is a supported endpoint such as predict that you want to call.

The inference API also supports asynchronous inference for long-running tasks and priority queuing.

​
Management API

Management API endpoints all run through the base api.baseten.co subdomain. Use management API endpoints for monitoring, CI/CD, and building both model-level and workspace-level automations.

Production deployment