Clone an endpoint to start a new one from the same settings, for example to try another model or machine while the original keeps running.
Before you begin
Section titled “Before you begin”- Permission to update and create model endpoints in the project. A Member needs a policy that grants update and create permission for the Inference service. Clone Inference is available with the update permission, and Clone Endpoint needs the create permission as well.
- Credit and free capacity for a new endpoint, as for any create.
Clone the endpoint
Section titled “Clone the endpoint”- In Model Endpoints, find the endpoint. Under Actions, or in the Actions menu on its details page, select Clone Inference. It is available for every endpoint.
- The create form opens, filled from the endpoint. The name starts with
cloned-, followed by the first part of the original name and a number made from the date and time. The workers are set to 1. Change the name before you deploy. - Change anything else you want, such as the model or the machine. See Create an endpoint.
- Select Clone Endpoint.
Check the clone
Section titled “Check the clone”The clone is a new endpoint. It opens on its details page and starts in Deploying, and the original stays as it was. If the clone cannot be created, see Troubleshooting Model Endpoints.
To change the original instead, see Update an endpoint.