Skip to content

Bria Attribution Service API Reference (1.2.0)

The Bria Attribution Service connects outputs of Bria models to the licensed training data behind them, so content owners are paid for their contribution. See Attribution Agent and Service for when you need it.

Assets you generate through Bria's hosted API are attributed automatically. Call this service when a Bria model runs somewhere else: in your own cloud, on a third-party inference platform, or from Bria's published weights.

The service supports images and video, each in two modes:

  • Send the asset (by_image, by_video): Bria computes the attribution. No compute on your side.
  • Send embeddings (by_embedding): the attribution model runs in your environment and only irreversible vectors reach Bria. See the bria-attribution repository.

Every request names the Bria model that produced the asset in model, written as model_name:version (for example image_fibo:2). Requests are asynchronous by default and return a request_id and status_url; track them with the status endpoint, or send sync: true to get the outcome in the response.

A COMPLETED status means Bria accepted the asset or embedding for attribution. The attribution itself is computed afterwards and is not returned in the response.

Languages
Servers
Production
https://engine.prod.bria-api.com