Implement generative AI solutions →hardMultiple Choice

AI-102 Implement generative AI solutions Practice Question

Exhibit

Refer to the exhibit.

```json
{
  "$schema": "https://schema.management.azure.com/schemas/2019-04-01/deploymentTemplate.json#",
  "contentVersion": "1.0.0.0",
  "resources": [
    {
      "type": "Microsoft.CognitiveServices/accounts",
      "apiVersion": "2023-05-01",
      "name": "myOpenAI",
      "location": "eastus",
      "sku": {
        "name": "S0"
      },
      "kind": "OpenAI",
      "properties": {
        "customSubDomainName": "myopenai"
      }
    },
    {
      "type": "Microsoft.CognitiveServices/accounts/deployments",
      "apiVersion": "2023-05-01",
      "name": "myOpenAI/gpt-35-turbo",
      "dependsOn": [
        "[resourceId('Microsoft.CognitiveServices/accounts', 'myOpenAI')]"
      ],
      "sku": {
        "name": "Standard",
        "capacity": 100
      },
      "properties": {
        "model": {
          "format": "OpenAI",
          "name": "gpt-35-turbo",
          "version": "0613"
        },
        "raiPolicyName": "MyPolicy"
      }
    }
  ]
}
```

You are reviewing an ARM template for deploying Azure OpenAI Service. The template includes a deployment for gpt-35-turbo with a capacity of 100. You need to ensure that the deployment uses provisioned throughput instead of standard. What should you modify?

⚠ Common exam trap

It's easy for candidates to think increasing capacity or changing the model version enables provisioned throughput, but the exam tests the specific SKU name 'ProvisionedManaged' as the only way to switch from standard to provisioned throughput in an ARM template.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Change the sku name to 'ProvisionedManaged'.

To use provisioned throughput (PTU) with Azure OpenAI Service, you must set the SKU name to 'ProvisionedManaged' in the ARM template. The default SKU is 'Standard', which uses pay-per-token consumption. Changing the SKU name to 'ProvisionedManaged' tells the resource provider to allocate dedicated throughput capacity for the deployment, ensuring consistent latency and throughput regardless of other workloads.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✓
Change the sku name to 'ProvisionedManaged'.
Why this is correct
Provisioned throughput requires the deployment's SKU name to be 'ProvisionedManaged' rather than 'Standard'; capacity then represents provisioned throughput units. Changing the sku name satisfies the stem's requirement to switch from standard to provisioned throughput deployment.
✗
Remove the raiPolicyName property.
Why it's wrong here
raiPolicyName configures the responsible AI content filter, unrelated to billing mode; removing it changes filtering, not throughput. Provisioned throughput is selected through the sku name (for example GlobalProvisionedManaged) and capacity units. The property is tempting because it appears in the deployment block, but it governs content policy only.
✗
Increase the capacity to 200.
Why it's wrong here
Capacity sets the provisioned throughput units (PTUs) only once the sku is 'Provisioned'; raising it to 200 while the sku remains 'Standard' still yields pay-as-you-go token billing. Higher capacity is tempting because provisioned deployments do consume capacity units, but the sku property is the actual switch.
✗
Change the model format to 'GPT-4'.
Why it's wrong here
Changing the model format to 'GPT-4' alters which model is deployed, not the billing or throughput mode. Provisioned throughput requires setting the deployment's sku to 'Provisioned' with a provisioned capacity. GPT-4 is tempting because it is a newer, more capable model, but model choice is orthogonal to throughput type.

Go deeper

AI-102 Implement generative AI solutions Practice Question

Exhibit

Why each option matters

Option-by-option breakdown

Related to this question