Keep seeing this message when deploying any GPT models: "Our system has detected this request as unusual activity for your account. If you are confident this is in error, please contact support."

Hello @Yuchen Wang

This error is generally not a GPT model quota, capacity, RBAC, or deployment-configuration issue. Similar Azure AI Foundry cases report error 715-123420, where an account- or subscription-level protection or risk-control mechanism rejects deployment requests before normal model provisioning occurs.

Since you're seeing the same message when deploying any GPT model, I wouldn't recommend repeatedly changing models, regions, SKUs, or recreating Foundry resources. Similar cases show that these changes don't clear the restriction.

First, verify that:

• The Azure subscription and billing profile are active.
• You're not connecting through a VPN/proxy when attempting the deployment.
• The required model quota is available.
• A single deployment attempt from the Azure portal produces the same error.

If it still fails, this generally requires Microsoft-side review. Open an Azure support request and provide the exact error, UTC timestamp, model/region, subscription and resource information, plus any Trace ID, Client Request ID, Service Request ID, or correlation ID shown by Foundry. Keep subscription/tenant IDs and other account details inside the private support case rather than posting them publicly.

If your subscription only has Basic/Free support and doesn't allow you to create a technical case, mention that explicitly here. Other recent Q&A cases show the same restriction, where users also couldn't create a technical support request.

Unfortunately, you can't normally clear this through a local Foundry configuration change. Microsoft needs to review the account/subscription-side restriction and determine whether the detection was a false positive.

Please "Accept the Answer" if this information helped you. This will help us and others in the community.

Hello @Yuchen Wang ,

Welcome to Microsoft Q&A .Thank you for reaching out to us.

The current behavior suggests that the deployment request may be failing during an earlier validation stage before the model provisioning process begins. The error message itself does not expose the exact validation reason, so confirmation requires additional backend investigation.

Possible areas that may require review include:

• Subscription eligibility validation
• Subscription-level authorization checks
• Automated risk or trust validation controls
• Model deployment access validation

Regarding the request for an RTFP review, the available information does not conclusively confirm that an RTFP restriction is present. However, the observed pattern is consistent with scenarios where additional backend validation may be required to determine whether a subscription-level condition is preventing deployments from completing.

The following checks can help rule out common deployment blockers before backend investigation.

1. Confirming Model Availability Verify that the selected models are supported in the intended region and deployment type. Checks:
   • Confirm model availability for the selected Azure region.
   • Confirm that the selected deployment type (for example, Global Standard) is supported for the model.
2. Validating Quota and Capacity Availability Confirm that sufficient quota is available for:
   • Subscription
   • Region
   • Selected model deployment
   Quota or capacity issues normally return more specific quota-related messages. However, validating quota helps eliminate standard deployment limitations.
3. Verifying Resource Provider Registration Confirm that the required resource providers are registered and in a healthy state:
   • Microsoft.CognitiveServices
   • Microsoft.MachineLearningServices
   Expected status: Registered
4. Reviewing Governance and Policy Restrictions Review Azure Policy assignments that may restrict:
   • AI model deployments
   • Allowed regions
   • Resource types
   • Deployment configurations
5. Performing Additional Deployment Validation To determine whether the issue is limited to a specific deployment method or occurs at the service level:
   • Retry after a short interval to rule out temporary service behavior.
   • Attempt a deployment with a new deployment name to exclude deployment-specific inconsistencies.
   • Test deployment through Azure AI Foundry portal.
   • If possible, test through Azure CLI or ARM/Bicep to confirm whether the failure occurs at the service layer.

As an additional verification , please note that

For sponsored subscriptions, it may also be useful to confirm that the selected models appear under the appropriate model collection in Azure AI Foundry, as model eligibility and billing behavior can vary depending on the model source.

The following references might be helpful , please check them out

• Region availability for Foundry Models sold by Azure - Microsoft Foundry | Microsoft Learn
• Azure OpenAI in Microsoft Foundry Models Quotas and Limits - Microsoft Foundry | Microsoft Learn
• Azure resource providers and types - Azure Resource Manager | Microsoft Learn
• Overview of Azure Policy - Azure Policy | Microsoft Learn
• az cognitiveservices account deployment | Microsoft Learn
• Sponsorship Coverage for Microsoft Foundry Models | Microsoft Learn

We have reached out to you on private messenger for further assistance

Thank you