Azure OpenAI Response API logs with store enabled are not visible in Stored Completion in AI Foundry

[mjmustafaev]

I’m using the Azure OpenAI Response API with store: true, but the logs are not showing up under Stored Completions in AI Foundry

[leestott]

Key Requirements for Stored Completions to Show Up

1. API Version Compatibility

• Stored completions are only supported from 2024-10-01-preview onward.
• For full functionality, use 2025-02-01-preview or later.

2. API Type

• You must use the Chat Completions API (/chat/completions) — not the legacy /completions endpoint.

3. Deployment Type

• Supported across all deployment types: standard, global, datazone, and provisioned.

4. Required Parameters
Make sure your request includes:

store: true,
metadata: {
  "user": "your-identifier",
  "category": "your-tag"
}

The metadata field isn’t strictly required, but it helps with traceability and filtering in the Foundry UI.

5. Model Deployment Name

• Use the correct model name (e.g., "gpt-4o") that matches your deployed endpoint.

Debugging Tips

• Double-check the API version in your client config or REST call.
• Confirm you're hitting the Chat Completions endpoint, not the older Completions one.
• Inspect the response object — if store: true is accepted, you should see a completion_id or similar traceable ID.
• Wait a few minutes — logs may take time to propagate to the Foundry dashboard.
• Use the AI Foundry Metrics Dashboard to confirm if any completions are being ingested at all.

Here's a minimal working example to validate that store: true is functioning correctly with the Azure OpenAI Response API and that completions are being logged under Stored Completions in AI Foundry.

Sample Request: Azure OpenAI Completion with store: true

POST https://{your-resource-name}.openai.azure.com/openai/deployments/{deployment-id}/completions?api-version=2024-02-15-preview
Content-Type: application/json
api-key: {your-api-key}

{
  "prompt": "Translate the following to Hausa: 'Welcome to our platform.'",
  "max_tokens": 50,
  "temperature": 0.7,
  "top_p": 1.0,
  "frequency_penalty": 0.0,
  "presence_penalty": 0.0,
  "store": true
}

Validation Steps in AI Foundry

1. Confirm Deployment
   Ensure the deployment ID matches the one registered in AI Foundry. If it's a custom model (e.g. Phi-4 with LoRA), confirm it's properly registered under Models > Deployments.

2. Check Completion Logs
   Go to AI Foundry > Monitoring > Stored Completions. You should see a log entry with:

   • Timestamp
   • Prompt content
   • Model name
   • Completion text

3. Verify Storage Settings
   In the Azure portal, navigate to your OpenAI resource and confirm that logging and storage are enabled. If using a custom endpoint or proxy, ensure it forwards the store: true flag correctly.

Troubleshooting Tips

• If logs aren’t appearing:
  • Confirm you're using the preview API version (2024-02-15-preview or newer).
  • Ensure your Azure role has Reader or higher access to the OpenAI resource and AI Foundry workspace.
  • Check if the completion is being routed through a proxy or wrapper that strips the store flag.

Helpful Resources

• [Stored Completions Setup Guide](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/stored-completions)
• [Monitoring Azure OpenAI in AI Foundry](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/monitor-openai)

[mjmustafaev]

Hi @leestott, understood, thanks!