Provisioned throughput for Foundry Models in Azure Government

Azure Government supports model-independent, or fungible, provisioned throughput unit (PTU) quota and Azure Reservations for eligible models. This article is the Azure Government companion to Provisioned throughput for Foundry Models. It describes only the requirements that differ in Azure Government. Use the shared provisioned throughput guidance for all other concepts and procedures.

Azure Government differences

Area	Azure Government behavior
Model eligibility	gpt-5.1 and earlier models use commitment-based PTUs. Models released after gpt-5.1 use fungible PTU quota and Azure Reservations when those models are available for provisioned deployment in Azure Government.
PTU quota	PTU quota is model-independent for eligible models. You can use the same quota pool for supported models within the same Azure Government region and provisioned deployment type.
Payment options	Provisioned deployments for models released after gpt-5.1 are billed hourly. You can use an Azure Reservation to apply a term discount. Monthly PTU commitments support gpt-5.1 and earlier models only.

Important

Eligibility for fungible PTU quota and Azure Reservations doesn't indicate model or regional availability. Check availability before you plan a deployment.

An Azure Reservation provides a billing discount. It doesn't reserve model capacity. A one-month Azure Reservation isn't a legacy monthly PTU commitment.

Transition from monthly commitments

Before this update, Azure Government provisioned deployments used monthly PTU commitments. The new hourly and Azure Reservation payment model applies to models released after gpt-5.1. This update doesn't change the commitment-based payment model for gpt-5.1 and earlier models.

For an eligible deployment, the PTUs that you deploy generate hourly charges. You can apply an Azure Reservation discount to those charges. The PTU quota determines how many PTUs you can deploy, while the reservation determines how the deployed PTUs are discounted.

Request fungible PTU quota

Use the Azure Government quota request form to request both standard quota and fungible PTU quota.

Having PTU quota doesn't guarantee that model capacity is available. Check model and regional availability before you plan a deployment.

Continue with provisioned throughput guidance

After you review the Azure Government differences, use the shared guidance for common provisioned throughput tasks.

Task	Guidance
Understand PTUs, quota, and capacity	Provisioned throughput for Foundry Models
Estimate the PTUs for a workload	Determine PTU sizing for a workload
Understand hourly billing and Azure Reservations	Provisioned throughput billing and cost management
Check models and regions	Foundry Models sold by Azure in Azure Government
Compare deployment types	Deployment types for Microsoft Foundry Models in Azure Government
Review quota limits	Azure OpenAI quotas and limits in Azure Government