Azure content understanding is very unstable today

Question

Azure content understanding is very unstable today

Tomas Masiulis 0

Hi,

Starting today, Content Understanding Studio has been returning errors when calling analyzers. The issue occurs regardless of whether the analysers are called from the user interface or via the API. No changes have been made to the configuration.

User's image

{

"id": "dd606eef-a2a4-4d4e-ad0f-538a1b0e35ba",

"status": "Failed",

"error": {

    "code": "InternalServerError",

    "message": "An unexpected error occurred."

},

"result": {

    "analyzerId": "projectAnalyzer_1776748484926_373",

    "apiVersion": "2025-11-01",

    "createdAt": "2026-06-10T12:11:28Z",

    "warnings": [],

    "contents": []

}

}

0 comments

2 answers

Your answer

Answer 1

The error payload and the fact that both Content Understanding Studio and API calls fail with InternalServerError and no config changes strongly indicate a service-side or regional issue rather than a problem with the analyzer definition or client code.

Concrete steps to take:

Validate that model deployments and defaults are still configured
- Open Content Understanding Studio.
- Select the Settings gear → Add resource and confirm the correct Microsoft Foundry resource is connected.
- Ensure the option to configure model deployments shows valid mappings for required models (for example gpt-4.1, gpt-4.1-mini, text-embedding-3-large). If defaults were removed or changed, re-save them or enable auto-deployment for required models if no default deployment available and save.
- If using REST or code, verify that resource-level defaults are still set via PATCH /contentunderstanding/defaults or that modelDeployments are passed explicitly in the analyze request.
Confirm required models are deployed in Microsoft Foundry
- In Azure Portal, go to the Microsoft Foundry resource → Deployments.
- Ensure the models required by the analyzers in use are deployed (for example gpt-4.1, gpt-4.1-mini, text-embedding-3-large) and that deployment names match what is configured as defaults.
- If any required model deployment is missing or was deleted/renamed, redeploy it and reconfigure defaults.
Check for platform or regional incidents
- Since the behavior started “today” and affects both Studio and API, check Azure Status and the Service Health blade in the Azure portal for incidents affecting Azure Content Understanding or Microsoft Foundry in the region where the resource is deployed.
- If possible, create a test Content Understanding/Microsoft Foundry resource in another supported region and connect it in Content Understanding Studio to see if the same analyzer runs successfully there. A success in another region further confirms a regional service issue.
Enable SDK logging to capture correlation details
- When calling via the JavaScript or .NET client libraries, enable logging so that request IDs and additional diagnostics are captured. These are useful if escalation to support is required.
- JavaScript: set AZURE_LOG_LEVEL=info or call setLogLevel("info") from @azure/logger.
- .NET: use AzureEventSourceListener.CreateConsoleLogger() to capture Azure SDK diagnostics.
If the problem persists after the above checks
- With confirmed model deployments, valid defaults, and no local configuration changes, and given the InternalServerError from both Studio and API, treat this as a service-side regression.
- Use the operation id (dd606eef-a2a4-4d4e-ad0f-538a1b0e35ba) and any client-side correlation IDs from logs when opening a support request so the service team can trace the failing operations.

References:

Answer 2

Tomas Masiulis 0

The AI response wasn't helpful. I tested it in a different region, and the same issue persists.
I believe the issue is easily reproducible with any configuration.

0 comments