Share via

Preflight validation fails with error 715-123420 for CognitiveServices model deployments in UK South

Chris 0 Reputation points
2026-03-11T08:53:39.3166667+00:00

Since approximately 2026-03-10 23:00 UTC, all ARM preflight validation (az deployment group validate, az deployment group what-if, and az deployment group create) fails with error 715-123420 - An error occurred. Please reach out to support for additional assistance. when the template includes any Microsoft.CognitiveServices/accounts/deployments child resource targeting an existing account in UK South.

Last successful deployment was 2026-03-10 22:45–22:51 UTC (same template, same resources — completed successfully).

Azure OpenAI Service
Azure OpenAI Service

An Azure service that provides access to OpenAI’s GPT-3 models with enterprise capabilities.


2 answers

Sort by: Most helpful
  1. Karnam Venkata Rajeswari 1,560 Reputation points Microsoft External Staff Moderator
    2026-03-20T13:30:12.38+00:00

    Hello Chris,

    Welcome to Microsoft Q&A .Thank you for reaching out.

    Adding to the inputs provided by Deepanshu katara , please check if the following helps.

    The deployment failure with error code 715‑123420 occurs during ARM preflight validation, which runs before any resources are created. This error indicates that the platform is unable to validate a Cognitive Services model deployment child resource before execution begins. The issue is not caused by template syntax, permissions, or quota, as the same template may succeed in other regions or earlier time windows.

    Please check if the following troubleshooting steps help:

    1. Platform and regional validation - Please review Azure Service Health for any active incidents or advisories impacting Cognitive Services in the UK South region to confirm whether the issue is region‑specific by attempting a validation‑only deployment in a different supported region.
    2. Tooling and client verification - Ensure that the latest supported version of Azure CLI or ARM tooling is in use, as outdated clients can fail preflight validation.Then retry validating using az deployment group validate or what‑if to confirm the error occurs consistently.
    3. Deployment method isolation-Temporarily remove the Microsoft.CognitiveServices/accounts/deployments child resource from the ARM or Bicep template.Then deploy the Cognitive Services account first, then create the model deployment separately using Azure CLI or REST to bypass the failing validation path.
    4. Template and API alignment - Please verify that the ARM or Bicep template uses a supported and current API version for Microsoft.CognitiveServices/accounts and accounts/deployments.Then confirm the deployment model name, version, and SKU are supported in the target region.

     

    References:

    Find error codes - Azure Resource Manager | Microsoft Learn

    Microsoft.CognitiveServices/accounts - Bicep, ARM template & Terraform AzAPI reference | Microsoft Learn

    Microsoft.CognitiveServices/accounts/deployments - Bicep, ARM template & Terraform AzAPI reference | Microsoft Learn

    Troubleshoot Bicep file deployments - Azure Resource Manager | Microsoft Learn

    What is Azure Service Health? - Azure Service Health | Microsoft Learn

     

    Thank you!

    0 comments No comments

  2. Deepanshu katara 18,145 Reputation points MVP Volunteer Moderator
    2026-03-12T06:02:56.2833333+00:00

    Hello , Welcome to MS Q&A

    Deploy Model Deployments Separately via Azure CLI (Bypass ARM Template)

    Decouple the broken child resource from the ARM template temporarily:

    # Step 1: Deploy everything EXCEPT Microsoft.CognitiveServices/accounts/deployments
    # (remove the child resource blocks from your template temporarily)
    az deployment group create \
      --resource-group <rg> \
      --template-file template-no-model-deployments.json
    
    # Step 2: Create the model deployments directly via REST/CLI
    az cognitiveservices account deployment create \
      --name <openai-account-name> \
      --resource-group <rg> \
      --deployment-name <deployment-name> \
      --model-name gpt-4o \
      --model-version "2024-11-20" \
      --model-format OpenAI \
      --sku-capacity 10 \
      --sku-name "Standard"
    
    
    

    Pls check and Try and let me know result

    Thanks

    Deepanshu

    0 comments No comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.