no model is deployed when selecting an existing project but new model deployment

[therealjohn]

v0.1.31-preview
Visit the docs at https://aka.ms/azd-ai-agent-docs

Retrieving agent templates...
? Select a language: Python
? Select a starter template: Hello World agent (Responses, without a framework, Python)
Let's get your project initialized.

Initializing an app to run on Azure (azd init)

  (✓) Done: Initialized git repository
  (✓) Done: Downloading template code to: C:\Users\johmil\projects\azd-ai-agent-tests\model-deployment-2


Installing required extensions...
  (-) Skipped: Installing azure.ai.agents extension (version 0.1.31-preview already installed)

Downloading manifest from GitHub...
✓ Manifest validated successfully
Downloading files...
  .dockerignore
  .env.example
  Dockerfile
  README.md
  agent.manifest.yaml
  agent.yaml
  main.py
  requirements.txt
Downloaded to: src\hello-world-python-responses

? How would you like to deploy your agent?: Container (Docker)
? How would you like to configure model(s) for your agent?: Use existing model deployment(s) from a Foundry project
Select an Azure subscription to look up available models and provision your Foundry project resources.
? Select subscription: azd ai agent development (redacted)
? Select a Foundry project: ai-account-g4uvbu4lxidzu / ai-project-adc-test-1 (northcentralus)
Found 2 container registry connections:
? Select a container registry connection to use for this agent: acr-connection
Using Application Insights connection: appi-connection (/subscriptions/redacted/resourceGroups/rg-adc-test-1/providers/Microsoft.Insights/components/appi-g4uvbu4lxidzu)
In your Microsoft Foundry project, found 1 existing model deployment(s) matching your model gpt-4.1-mini.
? Select deployment: Create new model deployment
? Select a version for gpt-4.1-mini: 2025-04-14 (default) [up to 50000 quota available]
? Select a SKU for gpt-4.1-mini v2025-04-14: Standard [170 quota available]
? Enter deployment capacity for gpt-4.1-mini (Standard): 10
? Enter model deployment name for model 'gpt-4.1-mini' (defaults to model name): gpt-4.1-mini2
Model deployment details processed and injected into agent definition. Deployment details can also be found in the JSON formatted AI_PROJECT_DEPLOYMENTS environment variable.
? Select container resource allocation (CPU and Memory) for your agent. You can adjust these settings later in the azure.yaml file if needed.: 0.25 cores, 0.5Gi memory
Processed agent.yaml at src\hello-world-python-responses\agent.yaml

Added your agent as a service entry named 'hello-world-python-responses' under the file azure.yaml.
To deploy your agent, use azd deploy hello-world-python-responses.

AI agent definition added to your azd project successfully!
Update available: 1.24.3 -> 1.25.0 (Release notes)
To update, run `azd update`
❯❯ model-deployment-2 git:(main) 09:24 azd provision

Provisioning Azure resources (azd provision)
Provisioning Azure resources can take some time.

Subscription: azd ai agent development (redacted)
Location: North Central US

  You can view detailed progress in the Azure Portal:
  https://portal.azure.com/#view/HubsExtension/DeploymentDetailsBlade/~/overview/id/%2Fsubscriptions%2redacted%2Fproviders%2FMicrosoft.Resources%2Fdeployments%2Fmodel-deployment-2-dev-1778592260

  (✓) Done: Resource group: rg-adc-test-1 (1.966s)

SUCCESS: Your application was provisioned in Azure in 28 seconds.
You can view the resources created under the resource group rg-adc-test-1 in Azure Portal:
https://portal.azure.com/#@/resource/subscriptionsredacted/resourceGroups/rg-adc-test-1/overview
Update available: 1.24.3 -> 1.25.0 (Release notes)
To update, run `azd update`

I expected a new model deployment in this project for gpt-4.1-mini2, but nothing is created.

I suspect it's an issue in the Bicep template for the existingProject path: https://github.com/Azure-Samples/azd-ai-starter-basic/tree/main/infra