Error 715-123420 blocks all model deployments (Serverless API) in Azure AI Foundry

Question

Error 715-123420 blocks all model deployments (Serverless API) in Azure AI Foundry

Shkodran Abdullahu 0

I'm unable to deploy any model as Serverless API (Model-as-a-Service) in my Azure AI Foundry project — this affects every model I've tried so far, not just one specific model. All attempts fail 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.

The failure happens right away, before provisioning starts, which suggests a subscription-level block rather than a model-specific or config issue. I first hit this trying to deploy Phi-4-mini-instruct, which is a first-party Microsoft "Consumption Service" model, not a third-party Marketplace offer — so the standard MaaS troubleshooting guide (country availability, marketplace purchase eligibility, RBAC permissions) doesn't seem to apply here.

Based on similar reports I've found, this error code appears to indicate a block by Azure's Real-Time Fraud Protection (RTFP) system at the subscription level. I'm on a Free/new Azure subscription. Requesting review and unblock.

Subscription ID: 035e200b-cc78-4a9c-xxxxxx
Resource group: xxxxx
Foundry resource: xxxxxxxx-foundry-resource

Shalini R 0 Reputation points

2026-07-23T13:00:07.7466667+00:00

This error is typically associated with Azure's Real-Time Fraud Protection (RTFP) system rather than the model itself. Since it occurs immediately for every Serverless API deployment, it does sound like a subscription-level restriction. Unfortunately, this isn't something you can resolve from the Foundry side—you'll likely need Azure Support to review and remove the block if it's a false positive.
Pavan Kumar Purilla 5 Reputation points Microsoft External Staff Moderator

2026-07-24T03:51:03.6166667+00:00

Hello Shkodran Abdullahu,

Thank you for reaching out to Microsoft Q&A.

When this error occurs (Error Code 715-123420 or 555420), it indicates that Azure AI Foundry has blocked GPT model deployment due to a fraud signal, and the request requires review by the appropriate internal team.

To proceed with further investigation and escalation, could you please share the details requested in private message for the same.

Thankyou!

1 answer

Your answer

Shalini R 0 Reputation points

2026-07-23T13:00:07.7466667+00:00

This error is typically associated with Azure's Real-Time Fraud Protection (RTFP) system rather than the model itself. Since it occurs immediately for every Serverless API deployment, it does sound like a subscription-level restriction. Unfortunately, this isn't something you can resolve from the Foundry side—you'll likely need Azure Support to review and remove the block if it's a false positive.
Pavan Kumar Purilla 5 Reputation points Microsoft External Staff Moderator

2026-07-24T03:51:03.6166667+00:00

Hello Shkodran Abdullahu,

Thank you for reaching out to Microsoft Q&A.

When this error occurs (Error Code 715-123420 or 555420), it indicates that Azure AI Foundry has blocked GPT model deployment due to a fraud signal, and the request requires review by the appropriate internal team.

To proceed with further investigation and escalation, could you please share the details requested in private message for the same.

Thankyou!

Answer 1

This does not appear to be a model-specific issue. The important detail is that the subscription is a Free or trial subscription. Microsoft’s current requirements for serverless API deployments state that the subscription must have a valid payment method and that Free or trial subscriptions are not supported, including for models sold directly by Azure. Therefore, the failure with Phi-4-mini-instruct does not prove that Marketplace permissions or model availability are the cause.

Upgrade the subscription to Pay-As-You-Go or use another paid subscription, confirm that the Foundry resource and project are in a supported region, and retry the deployment once. If error 715-123420 still occurs on the paid subscription across multiple first-party models, open an Azure support request and include the UTC failure time, region, resource name, Activity Log correlation ID, and error code.

Also remove the full subscription ID from the public post.