Question 13
Domain 1: Plan and Manage an Azure AI SolutionA data scientist at your company has been running Azure OpenAI experiments. The Azure Cost Management dashboard shows unexpected spend of $15,000 this month — 10x the budget. No usage alerts were configured. What should you implement to prevent this from happening again?
Correct answer: B
Explanation
Azure OpenAI usage is billed by token consumption, so setting token-per-minute (TPM) rate limits helps cap how much can be consumed in a given time. Azure Cost Management budget alerts provide notification when spending approaches or exceeds the budget, which addresses the missing usage alerts and helps prevent another $15,000 overrun.
Why each option is right or wrong
A. Restrict access to the Azure portal for all users
B. Configure token-per-minute (TPM) rate limits on the Azure OpenAI deployment and set Azure Cost Management budget alerts
Azure OpenAI deployments are governed by per-deployment throughput controls, including token-per-minute limits, which cap consumption at the service level and help prevent runaway prompt/completion usage from accumulating charges. In parallel, Azure Cost Management budgets can be configured with alert thresholds at 50%, 80%, 90%, and 100% of budget, so adding those alerts would have flagged the overspend before it reached $15,000.
C. Delete the Azure OpenAI resource and recreate it each month
D. Enable content filters to reduce the number of API calls