Error 715-123420 (RTFP fraud block) deploying gpt-5.5 and gpt-image-2 — request security-team review/unblock

Error 715-123420 (RTFP fraud block) isn’t described in the provided documentation, but there are a few supported checks and mitigations relevant to deploying GPT-5.5 and other Foundry models:

1. Confirm model access and region support
   • GPT-5.5 has specific access and quota requirements. Some quota tiers require explicit quota requests before deployment is allowed. Tier 5 and Tier 6 subscriptions have GPT-5.5 quota by default.
   • Verify that the target region supports GPT-5.5 and gpt-image-2 and that the subscription has access to these models in that region. If a model or version is not available in the selected region, deployment fails.
   • If region is the issue, switch to a region where the model is supported.
2. Check quota for the models
   • In Azure AI Foundry, go to Management Center → Quota and verify:
     • There is available Tokens-per-Minute (TPM) quota for GPT-5.5 and gpt-image-2 in the chosen region.
     • No “quota exceeded” or “regional capacity unavailable” status is shown for these models.
   • If quota is exhausted, either:
     • Reallocate TPM from existing deployments of the same model in that region, or
     • Submit a quota increase request using the quota increase form referenced in the documentation.
3. Validate deployment failures against known issues
   • For GPT-4.1 series there are known context/tool-size related failures, but these are runtime errors, not fraud blocks. This helps confirm that the 715-123420 error is not a standard context or payload issue but more likely related to access, policy, or backend validation.
4. Check for policy or access constraints
   • If there are Azure Policy assignments or security baselines restricting Azure AI Foundry usage (for example, policies around public network access or specific model families), these can cause deployment failures even when quota and region are correct.
   • Review subscription- and management-group–level policies that apply to Azure AI Foundry resources and confirm none are blocking GPT-5.x or image models.
5. When to escalate to support/security
   • If:
     • Region is supported,
     • Subscription tier and model access are correct,
     • Quota is available and not exceeded,
     • No blocking Azure Policy is in place,
   • and deployments still fail with the same fraud-related code, this indicates a backend validation or risk-flag condition that cannot be resolved from the portal.
   • In that case, open an Azure support request and include:
     • Subscription ID and region
     • Model IDs (gpt-5.5, gpt-image-2)
     • Exact error code and message (715-123420 RTFP fraud block)
     • Approximate timestamps and request IDs from failed deployment attempts
   • The support team can then route the case to the appropriate internal risk/security team to review and, if appropriate, remove or adjust the block on the subscription.

Summary actions:

1. Verify GPT-5.5 and gpt-image-2 are supported in the chosen region.
2. Confirm sufficient TPM quota for both models; request more or reallocate if needed.
3. Check for any Azure Policies or security controls that might be blocking these models.
4. If all checks pass and the error persists, raise an Azure support ticket with full error details so the backend/security team can review and unblock the subscription if warranted.

References:

• Foundry Models sold by Azure (azure-openai)
• Deploy Microsoft Foundry Models in the Foundry portal
• Azure OpenAI models and regions for Foundry Agent Service (classic)
• Role-based access control for Azure OpenAI in Microsoft Foundry Models (classic)
• Quota shows as fully utilised for Australia East region for Standard deployment - Microsoft Q&A