Region availability for models in serverless API endpoints | Azure Machine Learning

  • Article

In this article, you learn about which regions are available for each of the models supporting serverless API endpoint deployments.

Certain models in the model catalog can be deployed as a serverless API with pay-as-you-go billing. This kind of deployment provides a way to consume models as an API without hosting them on your subscription, while keeping the enterprise security and compliance that organizations need. This deployment option doesn't require quota from your subscription.

Region availability

Availability of serverless API endpoints for select models are listed in the following tables:

Cohere models

Region Cohere Command R Cohere Command R+ Cohere Embed v3
East US ✓ ✓ ✓
East US 2 ✓ ✓ ✓
Sweden Central ✓ ✓ ✓
North Central US ✓ ✓ ✓
South Central US ✓ ✓ ✓
West US ✓ ✓ ✓
West US 3 ✓ ✓ ✓

Mistral models

Region Mistral-Small Mistral-Large
East US ✓ ✓
East US 2 ✓ ✓
North Central US ✓ ✓
South Central US ✓ ✓
Sweden Central ✓ ✓
West US ✓ ✓
West US 3 ✓ ✓

Meta Llama models

Region Llama-2 Llama-3
East US ✓ ✓
East US 2 ✓ ✓
North Central US ✓ ✓
South Central US ✓ ✓
Sweden Central unavailable ✓
West US ✓ ✓
West US 3 ✓ ✓

Nixtla TimeGEN-1 model

Region Nixtla TimeGEN-1
East US ✓
East US 2 ✓
North Central US ✓
South Central US ✓
Sweden Central ✓
West US ✓
West US 3 ✓

Phi 3 models

Region Phi-3-mini Phi-3-small Phi-3-medium
East US 2 ✓ ✓ ✓
Sweden Central ✓ ✓ ✓

Jais model

Region Jais 30B Chat
East US ✓
East US 2 ✓
North Central US ✓
South Central US ✓
Sweden Central ✓
West US ✓
West US 3 ✓

AI21 Labs model

Region AI21-Jamba-Instruct
East US ✓
East US 2 ✓
North Central US ✓
South Central US ✓
Sweden Central ✓
West US ✓
West US 3 ✓

Note

Models offered through the Azure Marketplace are available for purchase only on Microsoft Managed Countries, with exception of Cohere family of models, which is also available in Japan.

Alternatives to region availability

If most of your infrastructure is in a particular region and you want to take advantage of models available only as serverless API endpoints, you can create a workspace on the supported region and then consume the endpoint from another region.

Read Consume serverless API endpoints from a different workspace to learn how to configure an existing serverless API endpoint in a different workspace than the one where it was deployed.

Related content