Failed to deploy model gpt-6-astra 715-123420: Our system has detected this request as unusual activity for your account. If you are confident this is in error, please contact support. Trace ID : da65a5b5-3e8f-441c-8f47-498192dc66b6 Client request ID : 5a

TN 0 Reputation points
2026-10-30T19:09:54.8933333+00:00

Since about 17:40 UTC on 30 October 2026, every attempt to create or update a model deployment on our Azure OpenAI / AI Foundry resource fails immediately with:

715-123420: Our system has detected this request as unusual activity for your account. If you are confident this is in error, please contact support.

  • Subscription ID: 00dea9c7-5fe8-4154-9c81-087ae303ae44
  • Resource ID: /subscriptions/00dea9c7-5fe8-4154-9c81-087ae303ae44/resourceGroups/developer-group/providers/Microsoft.CognitiveServices/accounts/talha-mdw0ttmb-eastus2
  • Region: East US 2
  • Trace IDs from the portal:
    • 48ce1370-d911-4d64-8587-c5ffdbadc986
    • 76049e09-d6d6-482e-995f-76aed375ecf4
  • The error occurs on every deployment on the resource we've tried, including gpt-4.1, gpt-4.1-mini, gpt-5, gpt-5-mini, gpt-6.1-sol, gpt-4o-transcribe and gpt-4o-transcribe-diarize, and when attempting to create a new deployment.
  • It occurs both in the Azure AI Foundry portal and through the Azure CLI / ARM API.

Existing deployments still serve inference requests normally. Only deployment create/update operations are blocked.

Context: shortly before the block began, I made several legitimate configuration changes within a few minutes. I created a custom content-filter (RAI) policy and assigned it to the deployments listed above. I believe this burst of changes was flagged automatically. All changes were made by the subscription owner for our own production workload. No credentials have been shared, and we have not seen any unauthorised access.

Request:

  1. Please review the subscription and lift the hold on deployment operations. Community guidance suggests this may need routing to the Real-Time Fraud Protection (RTFP) / Fraud Protection and Security team.
  2. Please confirm when the block is removed. We are not retrying deployment changes in the meantime, so the activity pattern doesn't get worse.

We are on the Basic support plan, so we're submitting this as a billing / subscription-management request. 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

1 answer

Sort by: Newest
Most helpful Newest Oldest
  1. Hemant Bharadwaj 355 Reputation points Microsoft Employee
    2026-11-01T14:50:28.5033333+00:00

    Likely cause: this specific 715-123420 message is typically a service-side risk/abuse-protection block on deployment requests (not a normal quota exhaustion error), so retries on the same request path usually keep failing until support reviews the block.

    1. Confirm basic deployment prerequisites in your target resource:
      • correct region/model availability
      • sufficient model quota/capacity for the target model
      • consistent subscription/tenant context
      References:
    2. Avoid repeated rapid retries with the same payload while blocked; this can prolong mitigation.
    3. Open a support request and provide the exact correlation details for backend unblock review:
    4. In parallel, validate that a minimal deployment request in the same resource/context still returns the same 715-123420 error (this helps confirm it is not payload-specific).

    Confirmed facts from your post:

    • Failures started around 2026-10-30 17:40 UTC.
    • Error is immediate and consistent across attempts.
    • You already captured trace/client request IDs.

    Possible cause (not yet confirmed): a false-positive unusual-activity protection flag at service level for the account/resource.

    Validation checks after remediation:

    • New deployment requests no longer return 715-123420.
    • Model deployment succeeds in the same region/resource.
    • Quota and deployment state are visible and healthy in Foundry/Azure portal.

    “If you find this answer helpful, please upvote it. If this resolves your issue, please mark it as the accepted answer so that it can help others in the community.”

    Was this answer helpful?

    0 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.