Skip to main content
The Frontier Gateway API manages the resources behind your branded inference gateway: the endpoints that route slugs to deployments, the groups that model your customers and their limits, and the federated API keys your customers call with. For the conceptual guide and end-to-end examples, see the Frontier Gateway overview. These endpoints live under /v1/gateway/, and Baseten enables them for Frontier Gateway workspaces. Authenticate with a workspace API key with management scope, passed as Authorization: Bearer $BASETEN_API_KEY. Requests from workspaces that aren’t onboarded return 403. The OpenAPI spec is available at api.baseten.co/v1/spec for generating API clients. It doesn’t yet cover deleting a group, revoking a group API key, or fetching group usage: see those reference pages for request and response details.

​
Endpoints

An endpoint maps a routing slug to the deployment that serves it. Use these routes to create, inspect, re-point, and delete endpoints.

​
Groups

A group owns the model set, rate limits, usage limits, and place in the hierarchy that its keys inherit. Use these routes to manage groups and inspect their usage.

​
API keys

A federated API key is the credential your customers call with, bound to one group whose limits it inherits. Use these routes to mint, list, and revoke keys.

​
Events

Every inference request through the gateway produces a billing event. Fetch events for a time window with this route, optionally filtered by API key prefix or external entity ID.

​
Billing webhooks

Instead of polling for events, Frontier Gateway can push a signed per-request usage event to your webhook endpoint for every call. For the payload shape and signature verification, see Billing webhooks.