Microsoft Foundry Model Evaluation and Serverless Deployment
Short answer
Configure Model Evaluation using the Model Catalog leaderboards and set the Deployment Option to Serverless.
Question
You have a Microsoft Foundry project for a support ticket triage app that reads ticket text from a database.
You need to ensure that the app meets the following requirements:
• Compares candidate models by quality, cost, and throughput
• Prevents hosting model weights in the Azure subscription
What should you configure? To answer, select the appropriate options in the answer area.
NOTE: Each correct selection is worth one point.
Key Takeaways
The question tests the distinction between local vs. managed hosting and the specific tools within Foundry for pre-deployment benchmarking (Model Catalog).
This scenario addresses configuring Microsoft Foundry for AI model evaluation and secure deployment, specifically focusing on comparing models via leaderboards and using serverless endpoints.
Common mistake: Learners often confuse 'Serverless' with 'Container Apps' or fail to identify that Model Catalog is the tool for comparison rather than just deployment.
Expert Analysis
Why the Answer Is Correct
To compare candidate models by quality, cost, and throughput, you must use the Model Catalog within Microsoft Foundry, which provides leaderboards and model cards for benchmarking. To prevent hosting model weights in your Azure subscription, you must select a Serverless deployment option, as this keeps the infrastructure and weights managed by the provider.
Why Other Candidate Answers Are Wrong
Options like 'Local Deployment' or 'Container Instances' would require managing the underlying compute and potentially storing weights locally, violating the security constraint. 'Managed Identity' is an authentication concept, not a deployment topology for preventing weight hosting.
Community Comment Notes
As user Amelia233 noted, "Model catalog leaderboards and model cards allow you to evaluate and compare models... Serverless API deployments" are the correct configuration for these specific requirements. The community consensus supports the use of Model Catalog for evaluation and Serverless for the deployment architecture.
Exam Strategy
When asked about 'comparing' models, look for evaluation tools like Leaderboards or Model Cards. When asked about 'not hosting weights' or 'managed infrastructure', select Serverless or Platform-as-a-Service options.
Frequently Asked Questions
What is the difference between Model Catalog and Model Registry?
Why does Serverless prevent hosting weights in my subscription?
Discussion 2 contributions
Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.
From public community threads 2
-
Amelia233 👍 1Correct Answers Model evaluation: Use model catalog leaderboards and model cards. Deployment option: Use a serverless deployment. Explanations Model evaluation: Model catalog leaderboards and model cards allow you to evaluate and compare models across benchmarks for quality, cost, latency, and throughput metrics prior to deployment. Deployment option: Serverless API deployments (Models-as-a-Service) host model weights in Microsoft-managed infrastructure, charging on a pay-as-you-go per-token basis and avoiding hosting/managing model weights or dedicated compute inside your own Azure subscription.
-
8ac3742 👍 1model leaderboard and model cards, serverless Model weights are the actual multi-gigabyte files that make up a machine learning model. Model Catalog Leaderboards: Directly rank and compare foundation models across industry benchmarks (quality/accuracy), latency/throughput metrics, and pricing tiers. Model Cards: Detail specific model architectures, intended use cases, context windows, limitations, and pricing breakdown.
Reader comments & corrections
No reader corrections yet — spotted an error or have a note? Share it below.
Related Analysis
Practice All AI-103 Questions
Access 88 questions with complete answers and detailed explanations.
View Full AI-103 Practice Test →