Error 715-123420 on Microsoft for Startups subscription — RTFP unblock request for Azure AI Foundry

Paul Marc 0 Reputation points
2026-11-05T19:19:26.02+00:00

Hello Microsoft Team,

We are a Microsoft for Startups company unable to deploy GPT-6 models on our Azure AI Foundry resource. Every new deployment fails immediately with HTTP 400 error 715-123420 ("Our system has detected this request as unusual activity for your account").

Our original support ticket (#2610290050000410) auto-posted to the Italian forum with only the generic "Azure" tag, so it was not routed to the Foundry team.

Environment & details:

  • Subscription ID: bb04d28f-3b8f-4b87-83e8-ffc384bbee26 (Microsoft for Startups, Tier 1)
  • Tenant ID: b00a6521-6cce-4032-b015-45b1f457f42f
  • Resource: ai-weltis-shared (Resource Group: rg-weltis-shared, Region: italynorth, Kind: AIServices, SKU: S0)
  • Models attempted: gpt-6.1-sol (2026-09-29) and gpt-6-astra (2026-09-03), DataZoneStandard
  • Quota status: 100K TPM DataZoneStandard available (0 used)
  • Existing deployment: gpt-5.4-mini deployed on 2026-10-26 on the same resource is running normally
  • Recent failed Correlation IDs:
    • ae388f16-163c-4411-a357-6acc6dd7ba26 (2026-10-29T06:34:32Z)
    • 88404724-e008-4615-9997-5874824ba6e7 (2026-10-29T06:34:35Z)
    • cdf69265-c87e-4f51-bcfc-0e5747bae6d3 (2026-10-29T07:14:20Z)

Could a Microsoft moderator please escalate this to the RTFP / Fraud Protection backend team to lift the deployment hold on our subscription? Happy to reply to a private message if anything else is needed.

Azure OpenAI in Foundry Models

1 answer

Sort by: Most helpful
  1. AI answer

    2026-11-05T19:20:51.93+00:00

    Open an Azure Subscription management support request asking for a 715-123420 account-level fraud/security review and routing to the team that can remove the deployment hold; this backend restriction cannot be cleared through quota or Foundry configuration changes.

    Error 715-123420 with “unusual activity” is handled as a fraud/risk signal that requires Microsoft’s service or security team to review the affected subscription. Closely matching cases were resolved after Support created an internal security collaboration and the backend block was remediated.^1^

    Submit the correctly routed request

    Because the previous ticket was routed to a generic forum, create a new request rather than relying on that forum post:

    1. In the Azure portal, select ? → Help + support → Create a support request.
    2. Select the affected subscription.
    3. If a Technical request is available under your startup benefit, select the service most closely related to Azure AI Foundry / Azure OpenAI in Foundry Models.
    4. If Technical support is unavailable, select Subscription management. Azure provides subscription-management support for every subscription; ask the agent to route the account-level restriction for fraud/security review. Technical requests otherwise depend on the attached support plan.^2^
    5. Use a title such as: Azure AI Foundry deployment blocked — 715-123420 — request fraud/security backend review
    6. Include:
      • Full error code and message
      • Subscription type: Microsoft for Startups Tier 1
      • Resource name, resource group, region, kind, and SKU
      • Models, versions, and deployment type attempted
      • UTC failure timestamps and correlation IDs
      • Confirmation that available DataZoneStandard quota is unused
      • Confirmation that an existing deployment works on the same resource
      • Previous ticket number and an explanation that it was misrouted to Microsoft Q&A
      • Explicit request:
        “Please create the required internal collaboration with the Security/RTFP service team to review and remove the 715-123420 deployment hold.”

    Select Return to support request → Next if the portal first presents recommended solutions. Complete Problem details, choose whether to permit advanced diagnostic collection, select the available support method, and submit.^2^

    The portal requires Owner, Contributor, Support Request Contributor, or equivalent Microsoft.Support/* permission at subscription scope to create the request.^2^

    If the Azure request is blocked or auto-closed

    Use the Microsoft for Startups support route:

    1. Open the Microsoft for Startups program page in Azure Portal.
    2. Select ? → Help + support → Create a support request.
    3. Attach the deployment error screenshot and reference the incorrectly routed ticket.
    4. Request Azure technical routing for the backend security review.^3^

    For startups using the older Startups Portal experience—applicable to organizations that joined before May 29, 2026—the documented path is ? → Submit a Support Ticket.^4^

    While the review is pending

    • Avoid repeated deployment attempts or creating replacement resources; these actions do not clear an account-level hold.
    • Do not post subscription IDs, tenant IDs, correlation IDs, or ticket details publicly again. Provide them only inside the authenticated support request or a moderator-initiated private exchange.
    • Do not provide API keys, access tokens, client secrets, or credentials.
    • Retry only after Support confirms that backend analysis or adjustments are complete. In a matching case, the guidance was to retry after 1–2 hours following adjustment, then wait up to 24 hours if it still failed.^5^

    The working deployment and unused quota are useful evidence for routing, but they do not remove this particular hold. Similar Foundry cases were restored only after internal review or remediation.^6^


    References

    1. 715-123420: An error occurred. Please reach out to support for additional assistance. - Microsoft Q&A
    2. Create an Azure support request
    3. Microsoft for Startups Program Page within the Azure Portal
    4. Microsoft for Startups Portal
    5. Azure AI Foundry deployment/update failing with error 715-123420. - Microsoft Q&A
    6. Unable to deploy any models in Microsoft Foundry (Error 715-123420) - Microsoft Q&A
    AI-generated content may be incorrect. Read our transparency notes for more information.

    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.