Regional deployments

Regional deployments restrict every replica in a deployment to a specific Baseten region. You can select the region through the Baseten dashboard, Baseten CLI, or Management API.

How placement works

Your organization’s compliance policy defines the regions where its deployments may run. A deployment region adds a narrower constraint for one deployment. It cannot expand the regions allowed by the compliance policy.

Deployment region	Compliance policy	Where replicas can run	Dashboard display
Selected	Absent or allows the selected region	The selected region only	The selected region
Selected	Excludes the selected region	The deployment request fails	No new deployment
Not selected	Present	Any region allowed by the policy	The allowed regions
Not selected	Absent	Any region in Baseten’s global infrastructure	Global

Global is the default when neither a deployment region nor a compliance policy limits placement. Baseten can place replicas in any region where the requested compute is available.

Regional deployment constraints

Keep the following in mind before selecting a region:

One region per deployment. All replicas run in the selected region. For multi-region fallbacks or region exclusion rules, use regional environments instead.
Region selection is not a capacity reservation. Baseten checks whether it can schedule the selected accelerator in the region under the active compliance policy. If it can’t, the deployment request fails. This check doesn’t reserve future capacity.
The region is fixed per deployment. You can’t change a deployment’s region in place. Changing regions creates a new deployment. See change or promote a regional deployment.
GPU deployments only. Regional selection is supported for deployments on GPU instance types. Deployments on shared CPU instance types can’t select a region.
Models only. Regional selection applies to model deployments. Chains and training jobs can’t select a region.

Regional endpoints

Regional deployments require a regional endpoint that enters the deployment’s region directly:

https://model-{model_id}-region-{region_slug}.api.baseten.co/predict

To copy the correct URL from the Baseten dashboard, open your model, select the deployment, and choose API endpoint. Regional endpoints support the same request paths as other deployment endpoints. For gRPC, use model-{model_id}-region-{region_slug}.grpc.api.baseten.co.

Baseten reserves environment names that start with region-, so a regional deployment endpoint like model-{model_id}-region-eu.api.baseten.co can never collide with a regional environment endpoint. If you try to create an environment with a reserved name, the request fails.

Available regions

The following regions are available for regional deployments.

Region slug	Location
`us`	United States
`eu`	European Union

Need a region that isn’t listed? Contact support.

Deploy to a region

Select a region with the Baseten CLI or REST API. In the dashboard, redeploy from the deployment settings to select or change a region.

Baseten CLI
REST API

To deploy a model to a region:Deploy from your model’s directory, passing a region slug from available regions:

baseten model push --region eu

For more information, see baseten model push.

To deploy a model to a region:Set the region field in the deployment payload you send to POST /v1/prepare_model_upload. baseten model push --region sends this field to the same endpoint.To check a deployment’s region:

curl https://api.baseten.co/v1/models/$MODEL_ID/deployments \
  -H "Authorization: Bearer $BASETEN_API_KEY"

{
  ...
  "region": {
    "slug": "eu",
    "display_name": "European Union"
  }
}

A null region means the deployment has no deployment-specific region. Baseten places replicas in any region where the required compute is available and your organization’s compliance policy permits. If no compliance policy applies, the dashboard displays Global.

Deploy errors

Region selection fails at deploy time with one of the following errors:

Error	Cause	What to do
`Region '...' cannot be found`	The region slug doesn’t exist or isn’t available to your organization.	Use a region from available regions.
`Instance type configuration does not support regional selection.`	The deployment’s instance type is a shared CPU type.	Deploy on a GPU instance type to select a region.
`Regional selection is only allowed on verified organizations. Contact support to get your organization verified.`	Your organization is unverified.	Get your organization verified to use regional deployments.
`Unable to schedule this deployment in region '...'. Please contact Baseten support for assistance.`	Baseten can’t schedule the requested configuration in the selected region.	Try another available region or contact support.
`Unable to schedule this deployment in region '...'. Your organization's compliance policy may conflict with the requested region. Please contact Baseten support for assistance.`	No available placement satisfies the selected region, deployment configuration, and active compliance policy.	Select a region allowed by the policy or contact support.
`This deployment's configuration cannot currently be scheduled. Please contact Baseten support for assistance.`	Baseten can’t schedule the combined deployment and compliance constraints.	Contact support.

List regions

To list the regions you can select, call the regions API:

curl https://api.baseten.co/v1/regions \
  -H "Authorization: Bearer $BASETEN_API_KEY"

[
  {
    "slug": "us",
    "display_name": "United States"
  },
  {
    "slug": "eu",
    "display_name": "European Union"
  }
]

The response lists only regions that your organization can select. If you use teams, call the same endpoint with your team ID to list the regions that team can select:

curl https://api.baseten.co/v1/teams/$TEAM_ID/regions \
  -H "Authorization: Bearer $BASETEN_API_KEY"

Change or promote a regional deployment

Because a deployment’s region is fixed, changing regions always creates a new deployment. The current deployment keeps serving until the new one is ready.

Change a deployment’s region

Dashboard
Baseten CLI

Open the deployment

Sign in to app.baseten.co, choose Dedicated Inference, select your model, and open the deployment under Deployments.

Open the region settings

In Deployment settings, under Instance type, choose Manage.

Select a region

In Region, set After redeployment to the new region. Select Global to remove the deployment-specific region constraint. Baseten still applies your organization’s compliance policy.

Redeploy

Choose Deploy with new region.

Push the model again with a different region:

baseten model push --region us

Promote a regional deployment

Promoting a regional deployment works like any other promotion, with two differences:

When you promote, you can keep the current region, pick a different one, or select Global to remove the deployment-specific region constraint. Baseten still applies your organization’s compliance policy.
A deployment with a selected region can’t be promoted into a regional environment. Such a promotion fails with A deployment with a specified region cannot target regional environment '{environment}'. To promote into a regional environment, first promote a deployment without a selected region.