Unable to deploy gpt-4.1 and text-embedding-3-large in Azure AI Foundry / Content Understanding setup (error 715-123420)

Error 715-123420 is a generic deployment failure. From the available guidance, the main supported causes and fixes for deployment failures of GPT and embedding models in Microsoft Foundry / Content Understanding are:

1. Verify that the models are actually available to the subscription and region
   • Some models might not be available for a given subscription or region even if they appear in the catalog.
   • Use the CLI to check which models are available to the Foundry (Cognitive Services) account in the chosen region:

          az extension add -n cognitiveservices   # if not already installed
          az cognitiveservices account list-models \
            --name <your-foundry-account-name> \
            --resource-group <your-rg>
     

   • If gpt-4.1 or text-embedding-3-large do not appear in the list for that account, they cannot be deployed there. In that case, create or use a Foundry resource in a region where those models are listed, and deploy there.
2. Check for quota or authorization issues
   • Common deployment failures are caused by:
     • Quota exceeded for the selected SKU or region.
     • Authorization failed because the identity lacks the required RBAC role.
   • For quota:
     • In the Foundry portal, check the quota for the resource and region and request an increase if the quota for the model/SKU is exhausted.
   • For authorization:
     • Ensure the identity used for deployment has at least Cognitive Services Contributor on the Foundry resource.
3. Ensure the Foundry resource is correctly configured for Content Understanding
   • Content Understanding requires a Microsoft Foundry resource created in a supported region and connected correctly:
     • Create the Foundry resource in a region listed in the Content Understanding region support documentation.
     • Assign Cognitive Services User to the user or service principal that configures model deployments.
   • Then deploy the required models from the Foundry portal:
     • gpt-4.1
     • gpt-4.1-mini
     • text-embedding-3-large
4. Use Content Understanding defaults or per-request model mappings
   • After the deployments succeed, configure Content Understanding to use them either by:
     • Setting default model deployments at the resource level (via PATCH /contentunderstanding/defaults or Content Understanding Studio), or
     • Passing modelDeployments in each analyze request.
5. If the models are listed as available and quota/permissions are correct but 715-123420 persists
   • This indicates a backend deployment issue that cannot be resolved from the client side. In that case, open an Azure support ticket for the Foundry resource and include:
     • Subscription ID, resource name, region
     • The exact model names (gpt-4.1, text-embedding-3-large)
     • The full CLI command and the error code 715-123420.
   • There is no indication in the available guidance that the account itself is “blocked”; the usual causes are region availability, quota, or authorization.

References:

• Deploy models using Azure CLI and Bicep (programming-language-bicep)
• How to troubleshoot your deployments and monitors in Microsoft Foundry portal (classic)
• Azure AI Content Understanding client library for JavaScript - version 1.0.0
• Azure AI Content Understanding client library for Python - version 1.0.1
• Model deployment options for Content Understanding analyzers
• Migrate from Azure Content Understanding Preview to GA
• What is a Content Understanding analyzer?