Error 715-123420 on every model deployment (Anthropic and Azure OpenAI), new pay-as-you-go subscription, support cases bounced twice

Question

Error 715-123420 on every model deployment (Anthropic and Azure OpenAI), new pay-as-you-go subscription, support cases bounced twice

Skene Bennellick 0

Since 22 September 2026 every model deployment on a new pay-as-you-go subscription fails with error 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").

What we have checked

Fails for Anthropic claude-opus-5 and claude-sonnet-5 (Foundry, GlobalStandard, East US 2 and West US 3) and for Azure OpenAI gpt-5.4-mini (GlobalStandard), so it is not model- or Marketplace-specific.
Fails from the portal and from the CLI (az cognitiveservices account deployment create), and for two different users (an Owner and a Contributor), so the block is at subscription level.
Quota is available and the Anthropic terms are accepted.
We have not deleted or recreated the Foundry resource.

Support requests so far

260922***: closed 29 Sep as a technical misroute, told to buy a Technical support plan.
260929***: Billing & Subscription Management replied 6 Oct that it is outside their scope and referred me to this forum.

The subscription has no paid support plan.

The owner is JC Logistics Limited (trading as Veridion Group), a New Zealand company, using the subscription for an internal logistics application.

Could a moderator please route this to the backend fraud protection team for review, as was done in question 5947131? Happy to provide any verification by private message.

0 comments

1 answer

Your answer

Answer 1

Divyesh Govaerdhanan 11,895 MVP Volunteer Moderator

HI Skene Bennellick,

Welcome to Microsoft Q&A,

Thanks for the detailed write-up. Error 715-123420 is consistent with a platform-level block from Azure’s fraud protection checks. It is applied before the request reaches model provisioning. That fits what you tested: it fails across models, regions, portal and CLI, and for both users. Only the backend team can review and release it, so nothing on your side will clear it.

I have escalated this thread for backend review. Meanwhile:

Please don’t delete or recreate the Foundry resource, and avoid repeated retries.
Watch for a private message and reply there with any verification details requested, not on this public thread.
Once the team confirms the block is cleared, retry the deployment after 1 to 2 hours. If it still fails, wait 24 hours and try again.

Please click Accept Answer and upvote if this helped.

Skene Bennellick 0 Reputation points

2026-10-15T19:40:32.99+00:00

Hi Divyesh, thank you for escalating this. Here's an update: I tried a single deployment today (16 October) and it still fails with 715-123420. I haven't had a private message so far. Can you tell whether the backend team has picked up the review? I'm happy to send any verification details by private message. As you advised, I haven't deleted or recreated the resource.