Azure AI Foundry gpt-5-mini deployment blocked with error 715-123420 on new subscription

elizabeth hahn 0 Reputation points
2026-08-25T22:45:12.93+00:00

I created a new Azure subscription and Microsoft Foundry project today for a personal development/portfolio project.

I am attempting to deploy gpt-5-mini using Global Standard. The model is available to my project, and the quota page shows 500K TPM available for gpt-5-mini Global Standard.

When I attempt to create the deployment through the Microsoft Foundry UI, it fails with:

HTTP 400 Bad Request Error code: 715-123420 Message: “Our system has detected this request as unusual activity for your account. If you are confident this is in error, please contact support.”

The Azure Activity Log confirms the failed operation as:

Operation: Microsoft.CognitiveServices/accounts/deployments/write Deployment: gpt-5-mini-1 Timestamp: August 25, 2026, approximately 5:48 PM EDT (21:48 UTC) Correlation ID: de919f16-79a1-4bfd-9cd0-f3a73e1d2c84

I attempted to create an Azure technical support request, but my subscription has the Basic support plan, which does not permit technical support requests without upgrading to a paid support plan.

I found another Microsoft Q&A case involving the same 715-123420 error where a Microsoft moderator requested the affected account information privately for review.

Could a Microsoft moderator please help route my subscription for review as a possible false-positive account/security restriction? I can provide the subscription ID, tenant ID, resource information, or any other account details needed via private message.

Thank you

Microsoft Foundry
Microsoft Foundry

A unified Azure platform for creating and managing AI models, agents, and applications with built‑in enterprise security, monitoring, and governance

0 comments No comments

2 answers

Sort by: Most helpful
  1. elizabeth hahn 0 Reputation points
    2026-08-28T15:05:08.8366667+00:00

    Update: This issue resolved without any configuration changes on my end.

    My original gpt-5-mini Global Standard deployment failed with error 715-123420 (“unusual activity”). I was also unable to open the recommended support ticket because my subscription does not include the required support plan. Attempting to create the ticket through the Azure CLI resulted in InvalidSupportPlan.

    I did not repeatedly retry the deployment. After waiting approximately three days, I attempted the same gpt-5-mini Global Standard deployment again using the same Azure AI Foundry resource. This time the deployment succeeded, and I confirmed that the model responds successfully in the playground.

    I did not change the resource, region, deployment type, quota settings, or subscription.

    So in my case, the 715-123420 restriction appears to have cleared on the Microsoft/Azure side after a few days.

    I hope this helps others encountering the same issue, particularly users on the Basic support plan who cannot create the support ticket that is commonly recommended for this error.

    Was this answer helpful?

    0 comments No comments

  2. Jerald Felix 18,760 Reputation points Volunteer Moderator
    2026-08-26T16:40:27.6666667+00:00

    Hello elizabeth hahn,

    Greetings! Thanks for raising this question in the Q&A forum.

    The error code 715-123420 with the message "Our system has detected this request as unusual activity for your account" is not a quota, RBAC, or model-configuration issue on your side. Your quota page correctly shows 500K TPM available, and that is not what is blocking the deployment. This error indicates that Azure's Real-Time Fraud Protection (RTFP) system has placed a platform-level block on the deployment request at the subscription level, before provisioning even begins. This is common on brand-new subscriptions, especially ones just created for personal or portfolio projects, since new subscriptions with limited billing history are more likely to trigger these automated risk checks. This is not something you can resolve through the Foundry UI, CLI, or by changing deployment settings, and it requires manual review by Microsoft's internal Fraud Protection and Security team.

    Here is how to proceed:

    Do not delete or recreate resources. Repeatedly deleting and recreating the resource, project, or deployment will not clear the flag and can make the activity pattern look more suspicious to the automated system, potentially prolonging the review.

    Open a Subscription Management support ticket, not a Technical support ticket. Billing and subscription management support, which includes account and security flag reviews, is included at no cost on every Azure support plan, including Basic. You do not need to upgrade your plan for this. Go to:

    https://portal.azure.com/#blade/Microsoft_Azure_Support/HelpAndSupportBlade
    

    Select New support request, then for Issue type choose Billing (or Subscription Management), not Technical. This path does not require a paid support plan.

    In the ticket description, include the following details so the backend reviewer can locate your specific event quickly:

    • Subscription ID
      • Tenant ID
        • Error code: 715-123420
          • Resource name and resource group for the AI Foundry project
            • Correlation ID: de919f16-79a1-4bfd-9cd0-f3a73e1d2c84
              • Operation: Microsoft.CognitiveServices/accounts/deployments/write
                • Timestamp: August 25, 2026, approximately 21:48 UTC
                  • A short note that this is believed to be an RTFP/fraud-detection false positive on a newly created subscription
                  Reference prior related cases if asked. You can mention that this error code has been reported by multiple users on Microsoft Q&A as tied to the RTFP validation system, which may help the reviewer route it faster internally.

    Typical turnaround for this kind of review has varied from a few days to a couple of weeks depending on backlog, so it is worth checking back on the ticket periodically rather than resubmitting.

    If this answer helps you kindly accept the answer which will help others who have similar questions.

    Best Regards,

    Jerald Felix.

    Was this answer helpful?


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.