Error 715-123420 blocks Azure OpenAI GPT-5 and Foundry model deployments despite visible quota

We are blocked from deploying GPT-5-family Azure OpenAI models and Azure AI Foundry / AIServices marketplace models. Deployment attempts fail immediately with:

715-123420: An error occurred. Please reach out to support for additional assistance.

This appears similar to other Microsoft Q&A reports where 715-123420 is described as a platform-level / RTFP / backend validation block rather than a normal quota, region, RBAC, or resource-configuration issue. We are requesting Microsoft moderator help to route this to the internal team that can review and unblock 715-123420 / RTFP / Fraud Protection / Security cases, or create a secure collaboration ticket if needed.

What works:

• Normal Azure resources work. We successfully created, logged into, and deleted an Azure VM.
• Microsoft.CognitiveServices provider is Registered.
• Azure OpenAI resource creation works.
• gpt-4.1-mini deployment works and API calls succeed.

Working Azure OpenAI resource:

/subscriptions/d973f401-0cab-424f-81e4-d61d854ea6a3/resourceGroups/rg-oai-smoke-20260521/providers/Microsoft.CognitiveServices/accounts/vp05210835oai28074

Working deployment:

• Model: gpt-4.1-mini
• Version: 2025-04-14
• SKU: Standard
• Region: westus
• Result: deployment succeeds and API calls work.

Failing Azure OpenAI deployments:

1. gpt-5.4-mini
   • Model/version: gpt-5.4-mini 2026-03-17
   • SKU tried: DataZoneStandard
   • Quota visible: OpenAI.DataZoneStandard.gpt-5.4-mini = 1000
   • Error: 715-123420
   • Correlation IDs:
     • 29134a9d-339d-46ed-a6fd-3afaf00a28f6
     • f0fd8d15-3409-404b-9f74-4d0630c93a62
2. gpt-5-mini
   • Model/version: gpt-5-mini 2025-08-07
   • SKU tried: GlobalStandard
   • Region tried: eastus2
   • Quota visible: OpenAI.GlobalStandard.gpt-5-mini = 500
   • Error: 715-123420
   • Correlation ID:
     • f83aa64d-1ae6-464c-a997-d6b9cac93d44

Failing Azure AI Foundry / AIServices marketplace deployments:

AIServices account creation succeeds, but model deployments fail with the same 715-123420 code.

1. Kimi-K2.6
   • Version: 2026-04-20
   • Format/provider: MoonshotAI
   • SKU: GlobalStandard
   • Region: westus2
   • Quota visible: 100
   • Error: 715-123420
   • Correlation ID:
     • 6c47c136-5acf-4319-bb14-6db9d88858e0
2. DeepSeek-V4-Pro
   • Version: 2026-04-23
   • Format/provider: DeepSeek
   • SKU: GlobalStandard
   • Region: westus2
   • Quota visible: 1000
   • Error: 715-123420
   • Correlation ID:
     • 7838c879-0835-4a1c-a6c6-fb0e25be8146
3. Kimi-K2.5
   • SKU: GlobalStandard
   • Region: westus2
   • Quota visible: 100
   • Error: 715-123420
   • Correlation ID:
     • 64989086-76aa-46a0-bf3b-c397753f2f8d

We also saw the same 715-123420 behavior with additional Foundry model attempts such as DeepSeek-V3.2 and gpt-oss-120b.

Why this does not look like normal configuration failure:

• A basic Azure VM lifecycle works.
• Azure OpenAI resource creation works.
• gpt-4.1-mini deploys and runs successfully.
• Quota is visible for the failing models.
• The failures happen across multiple model families and providers.
• The failures happen immediately during deployment.
• Recreating resources and trying other model/region combinations did not resolve it.

Basic Azure support blocks technical ticket creation for this issue. It only allows billing, subscription management, and quota requests. This issue appears technical/backend-policy related, not billing. If this cannot be handled through Microsoft Q&A / moderator routing, please let us know whether the Developer support plan is sufficient to get this reviewed. Our current plan is to upgrade only to Developer support if required; we are not planning to purchase Standard or Professional Direct support for this issue.

Request:

Please help route this to the correct Microsoft internal team for 715-123420 / RTFP / Fraud Protection / Security review, or let us know the exact support path required. We can provide any additional non-secret identifiers needed through a secure channel. We will not share API keys, secrets, or credentials.