Skip to main content
The Baseten target adjusts the autoscaling range of one model deployment. Baseten continues choosing the live replica count within that range.

​
Configuration

The model and deployment IDs are available in Baseten’s deployment details and API URLs.

​
Authentication

Create an HTTP Auth Headers credential with the header Baseten expects for its API key:
Keep the key in the credential and select that credential when creating the target.

​
Observations and changes

Arklow reads the deployment status and active replica count. A deployment in a transitional state, such as deploying or updating, is allowed to finish before another change is applied. Scale changes update Baseten’s minimum and maximum replicas together. The observed active replica count may take time to respond while Baseten provisions or removes replicas.

​
Add a Baseten target

1

Open scale targets

Go to Scale Targets, click New scale target, and select Baseten.
2

Identify the deployment

Enter the Model ID and Deployment ID.
3

Select authentication

Choose the HTTP Auth Headers credential containing your Baseten API key.
4

Select capacity

Choose the destination or shared pool served by this deployment.
5

Set the envelope

Enter the minimum and maximum replica counts available to the Baseten autoscaler.
6

Save

Create the target and confirm that Last observed advances without a credential or deployment error.