Deploy GPT 5.4 model

Question

Hi Team ,

Subscription - Removed [PII]

Resource group – Removed[PII] 

We need to deploy GPT-5.4 model it throws " Restricted by admin policy: This model and version cannot be deployed due to a policy in your subscription. Contact your admin to request access."

Need help to update the policy or suggest how to fix this issue.

Answer 1

Hi Annapurani Sundaram (EXT),

That message means your subscription has an Azure Policy assignment that is blocking GPT 5.4 deployments. This is not a model problem and not a quota problem. It is a governance rule that says which models can be deployed in your subscription. [clemenssiebler.com]

Here is how you can check and update the Azure Policy that is blocking the deployment.

Step 1. Find the policy that is denying the deployment Open the Azure portal and go to Policy. Then open Assignments. Filter to your subscription and, if needed, your resource group. Look for a policy assignment related to model deployment or approved models.

Step 2. Confirm you have the right permissions to change it To update a policy assignment, you typically need Owner or Resource Policy Contributor on the scope where the policy is assigned, such as the subscription or resource group.

Step 3. Edit the policy assignment to allow GPT 5.4 Open the relevant policy assignment and select Edit assignment. Go to Parameters and show all parameters. Then update the allowed model settings so GPT 5.4 is included. Some policies use an allowed publishers list and an allowed asset or model ID list. If GPT 5.4 is not in those lists, the deployment will continue to be blocked.

Step 4. Get the exact model asset ID for GPT 5.4 and add it Go to the Foundry model catalog in the Foundry portal, open the GPT 5.4 model details, and copy the model ID or asset ID shown there. The policy needs this value to match exactly. Add that model ID to the allowed asset IDs list in the policy parameters. Also make sure the publisher, for example OpenAI, is included in allowed publishers if that parameter exists.

Step 5. Check the policy effect Some policies are set to Deny, which blocks deployments. Others can be set to Audit, which only logs. If your goal is to deploy GPT 5.4, the policy must either allow GPT 5.4 under Deny or be changed to a non-blocking effect depending on how your org manages governance.

Step 6. Retry the deployment after saving After you save the updated policy assignment, retry deploying GPT 5.4 from the Foundry portal.

Now a quick summary of the key deployment restrictions for GPT 5.4 in this situation

First, a subscription level admin policy can block a model even if the model exists and you have permissions to deploy resources. That is exactly what your error message is indicating.

Second, model availability can vary by region and by what your Azure OpenAI resource can see as deployable. If a model does not appear as deployable in your region, you may not be able to deploy it there.

Official reference for how Azure OpenAI model availability and deployment updates work
Working with models in Azure OpenAI in Microsoft Foundry Models. [microsoftl....github.io]

I Hope this helps. Do let me know if you have any further queries.

Thankyou!