Failed to create Azure OpenAI deployment InternalServerError: Service temporarily unavailable. Please try again later

Hello Aleksandar Kavaev,

Welcome to the Microsoft Q&A and thank you for posting your questions here.

I understand that your Azure OpenAI deployment failed with InternalServerError: Service temporarily unavailable.

Why will it fail? Hmmmm. Follow the step by steps below and resolve the issue:

Step 1: Go to https://free.blessedness.top/en-us/azure/ai-foundry/openai/quotas-limits?tabs=REST

• Confirm the followings:
  • Max 5 fine-tuned deployments per resource.
  • Max 32 total deployments.
  • Use Azure CLI bash command as below:

        az cognitiveservices account list-usage --name <resource_name> --resource-group <resource_group>
    

Step 2: Verify region support for GPT-4.1 Fine-Tuned Hosting - https://free.blessedness.top/en-us/azure/ai-foundry/foundry-models/concepts/models-sold-directly-by-azure?tabs=global-standard-aoai%2Cstandard-chat-completions%2Cglobal-standard to ensure your region supports GPT-4.1 fine-tuned hosting and if not, switch to a supported region.

Step 3: Make sure you or user has:

• Cognitive Services OpenAI Contributor
• Check via Azure Portal > IAM > Role Assignments.

Step 4: Use Azure CLI or Portal to confirm deletion of the previous deployment.

• Wait 5–10 minutes after deletion before retrying.
• Use: az cognitiveservices account deployment list --name <resource_name> --resource-group <resource_group> via your Azure CLI bash command.

Step 5: Use the provided Python script by @Sridhar M with correct values:

• sku.name = "standard"
• capacity = 1
• model.name = "gpt-4.1-YYYY-MM-DD.ft-<model_id>"
• Ensure:
  • api-version = "2024-10-01"
  • Model format is "OpenAI" and version is "1"

Step 6: Go to Azure Portal > Monitor > Activity Log to look for failed deployment events and error codes. Copy for later use.

Step 7: Avoid preview types like:

• Global Standard
• PTU
• Use:
  • Standard Chat Completions

To retry with Standard Deployment Type.

I hope this is helpful! Do not hesitate to let me know if you have any other questions or clarifications.

Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.

Thank you for your guidance.

I went through the steps and at the end I get this error:
400 Bad Request {"error":{"code":"SpecialFeatureOrQuotaIdRequired","message":"The current subscription does not have feature required by this model 'Format:OpenAI,Name:gpt-4.1-2025-04-14.ft-xxxxxxxxx,Version:1' and SKU 'Standard' on location EASTUS."}}

Currently I have only 1 base model gpt-4.1 deployed with not full quota for TPM in EASTUS.

Hi there

Thanks for using QandA plaftform

you mentioned that base models deploy fine and only fine-tuned ones fail, it’s likely a service capacity or internal availability problem specific to fine-tuned deployments. Azure OpenAI sometimes experiences short-term outages or throttling when allocating compute for custom models, especially for GPT-4.1. I’ve seen similar cases where the issue resolved itself after a few hours once the backend stabilized. You can try re-deploying later, ensuring that you’re using the same region where your fine-tuned model was trained since cross-region deployments aren’t supported. It’s also a good idea to check the Azure Service Health dashboard or AI Foundry logs for any ongoing regional incidents. If it still fails after 24 hours, I’d recommend raising a support ticket with the timestamp, resource name, and region so the Azure team can manually verify if your model deployment got stuck.

If this helps kindly accept the answer thanks much.