AI-300 Practice Question 47: Bicep Second Model Deployment
Microsoft Machine Learning Operations Engineer Associate (AI-300) · Test 5 · question 7 of 10
A team manages its Foundry resource with Bicep. Development and production applications must call the same model, and the team wants separate deployments so that quota, monitoring and version changes can be handled independently. How should they define the two deployments in the Bicep template?
Show answer and explanation
Answer:
Model deployments are child resources of the Foundry account (Microsoft.CognitiveServices/accounts/deployments), and the deployment documentation says that *"you can deploy the same model multiple times if needed as long as you give it a different deployment name."* Two child resources with different names create two independent deployments, each with its own capacity settings, which is what the team needs for dev and prod (option C). The same can also be done from the Azure CLI, since the az cognitiveservices commands are part of the core CLI.
- A is wrong because there is no runtime alias feature that turns one deployment into two. Each deployment has its own name and settings. - B is wrong because an account can hold more than one deployment of the same model as long as the deployment names differ, and separate accounts add cost and management without being needed. - D is wrong because sharing one deployment means sharing one quota and configuration, so version changes and limits for dev would affect prod, which is what the team wants to avoid.
Reference: Deploy models using Azure CLI and Bicep - Microsoft Foundry