Deploy Open AI resource using Terraform

Question

I use Terraform by deploying an Azure OpenAI resource by the following code:

resource "azurerm_cognitive_account" "openai" {
  name                = var.openai_name
  location            = var.resource_location
  resource_group_name = var.resource_group_name

  kind     = "OpenAI"
  sku_name = "S0"
}

However, this code seems to deploy an Azure OpenAI resource with a region based endpoint instead of a resource based endpoint. See here the difference when I do it via the Azure portal vs via Terraform code:

From the portal

vs.

Using Terraform code

Why does this happen and how can I create an Azure OpenAI resource with a region based endpoint using Terraform?

Answer 1

Basically, the functionality of terraform resource azurerm_cognitive_account open AI supports a region-based endpoints deployment as shown in the below code.

data "azurerm_resource_group" "example" {  
   name = "xxx"  
}
resource "azurerm_cognitive_account" "openai" {  
   name = "regionalai"  
   location = data.azurerm_resource_group.example.location  
   resource_group_name = data.azurerm_resource_group.example.name
     kind = "OpenAI"  
     sku_name = "S0"  
}

Deployment succeeded:

If you want to explicitly specify an endpoint as resource-based or any other, you can use custom_subdomain_nameparameter available in azurerm_cognitive_account terraform resource as shown below.

data "azurerm_resource_group" "example" {
  name     = "xxxx"
}
resource "azurerm_cognitive_account" "openai" {
  name                = "regionalai"
  location            = data.azurerm_resource_group.example.location
  resource_group_name = data.azurerm_resource_group.example.name

  kind     = "OpenAI"
  sku_name = "S0"
  custom_subdomain_name = "testvkdomain"
}

Output: