Unable to change token per minute limit even if there is enough in the deployment model.

The TPM change is being rejected by model-governance validation—not by available quota—because the update fails with InvalidResourceProperties: Failed to validate policies for model gpt-5-mini/2025-08-07.

Check the governing Azure Policy

Foundry’s built-in policies are evaluated at deployment time and can restrict deployments by approved model/publisher, source, or lifecycle status. Available TPM does not override those policies.^1^

1. In the Azure portal, open Policy.
2. Review policy assignments applying to the deployment’s:
   • Resource group
   • Subscription
   • Management group
3. Look specifically for:
   • Foundry model deployments should only use approved models
   • Foundry model deployments should meet eligibility requirements
4. Open the assignment and inspect its parameters:
   • For approved models, ensure the allowed publishers include OpenAI, or that the allowed asset IDs include the exact gpt-5-mini model/version.
   • For eligibility requirements, check whether source or lifecycle restrictions exclude this model.
5. If the policy is intentional, ask an Owner or Resource Policy Contributor at the applicable scope to update its allow-list or create an approved exemption. Those permissions are required to create or assign these policies.^1^
6. Retry the TPM update after the policy change has taken effect.

Do not remove an organization-wide policy solely to complete this update; modify the allow-list or use an exemption approved by the policy owner.

Change the TPM allocation after policy validation succeeds

For a standard deployment in the current Foundry portal:

1. Open Microsoft Foundry and select the correct project.
2. Select Manage → Quota → Token per minute.
3. Select the relevant model quota entry.
4. Under Affiliated deployments using shared quota, select the pencil icon beside gpt-5.0-mini-agentic-ai-prod-us.
5. Enter the new allocation and save.
6. Allow up to 15 minutes, refresh the quota page, and verify the allocation.

Quota is scoped per subscription, region, model, and deployment type. The sum of allocations to deployments sharing that quota cannot exceed its approved total. If necessary, reduce an unused deployment’s allocation or select Request quota.^2^

TPM also cannot be set to an arbitrary value: deployment TPM is modified in increments of 1,000. Changing TPM adjusts both the TPM limit and the corresponding RPM limit.^3^

Also note that deployment TPM is a throughput allocation; it does not change the model’s maximum input-token limit.^2^

If no applicable policy explains the failure

Capture the complete failure from More details, including:

• Exact error text
• Trace ID
• Client request ID
• Service request ID
• UTC timestamp
• Subscription and resource names
• Region, deployment type, model, and version
• Current and requested TPM values
• Relevant policy assignments and compliance results

Then open the Azure OpenAI resource in the Azure portal and select Support + Troubleshooting → Create a support request. Creating the request requires Owner, Contributor, Support Request Contributor, or equivalent Microsoft.Support/* permission at subscription scope.^4^

References

1. Built-in policies for model deployment - Microsoft Foundry
2. Manage Azure OpenAI in Microsoft Foundry Models quota
3. Manage Azure OpenAI in Azure AI Foundry Models quota
4. Create an Azure support request