# Missing reserved PTUs for steady-state Azure OpenAI workloads Canonical: https://www.pointfive.co/efficiency-hub/inefficiencies/missing-reserved-ptus-for-steady-state-azure-openai-workloads-f78d4 Many production Azure OpenAI workloads - such as chatbots, inference services, and retrieval-augmented generation (RAG) pipelines-use PTUs consistently... By: Ariel Fishman-Lichterman Updated: 2026-09-28 [Cloud Efficiency Hub](https://www.pointfive.co/efficiency-hub) # Missing Reserved PTUs for Steady-State Azure OpenAI Workloads The short version Many production Azure OpenAI workloads - such as chatbots, inference services, and retrieval-augmented generation (RAG) pipelines-use PTUs consistently throughout the day. [Ariel Fishman-Lichterman](https://www.linkedin.com/in/ariel-fishman-lichterman-01a096214/) Cloud Researcher @ PointFive Azure service [Azure Cognitive Services](https://www.pointfive.co/efficiency-hub/cloud-services/azure-cognitive-services) Category [AI](https://www.pointfive.co/efficiency-hub/service-category/ai) Reference CER-0246 Type Suboptimal Pricing Model ## Explanation Why the waste happens and who it affects. When usage stabilizes after initial experimentation, continuing to rely on on-demand PTUs results in ongoing unnecessary spend. These workloads are strong candidates for reserved PTUs, which provide identical performance guarantees at a substantially reduced hourly rate. Migrating to reservations requires no architectural changes, because a reservation is a billing discount applied to matching deployments. It must match the deployment type (and, for Data Zone and Regional deployments, the region), does not guarantee capacity, and is not canceled automatically when a deployment is deleted. ## Billing model The pricing dimensions that drive this cost. PTUs are billed hourly based on the number of PTUs deployed, whether or not the deployment is handling requests. On-demand PTUs use standard hourly rates, whereas reservations apply a discounted effective hourly rate when capacity is committed for one month or one year. The discount varies by deployment type (Global, Data Zone, or Regional), model family, and term, so compare current hourly and reservation prices before committing. Workloads running continuously on on-demand PTUs incur avoidable premium pricing. ## How to detect 4 checks to find it in your estate. - Review PTU deployments supporting production workloads that operate continuously throughout the day - Assess whether throughput demand remains stable enough to justify reserved capacity - Identify deployments that have moved beyond experimentation but still use on-demand PTUs - Evaluate the cost difference between on-demand PTUs and reserved PTUs for these workloads ## How to fix 4 ways to remove the waste. - Purchase monthly or annual reserved PTUs for workloads with sustained, predictable throughput needs, buying them only after the deployments exist and matching each reservation to the deployment type and region - Establish governance criteria defining when production PTU deployments should transition to reservations - Periodically reassess workload stability to ensure PTU reservation commitments remain aligned with demand - Use cost modeling to evaluate reservation options as part of production readiness reviews ## Documentation Vendor references for pricing and configuration. - [Save costs with Microsoft Foundry Provisioned Throughput Reservations learn.microsoft.com](https://learn.microsoft.com/en-us/azure/cost-management-billing/reservations/microsoft-foundry) - [Provisioned throughput billing and cost management learn.microsoft.com](https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/provisioned-throughput-billing) ## Related inefficiencies [Browse the library](https://www.pointfive.co/efficiency-hub) - Azure Cognitive Services CER-0248 ### [Non-Production Azure OpenAI Deployments Using PTUs Instead of PAYG](https://www.pointfive.co/efficiency-hub/inefficiencies/non-production-azure-openai-deployments-using-ptus-instead-of-payg-9fe81) Development, testing, QA, and sandbox environments rarely have the steady, predictable traffic patterns needed to justify PTU deployments. These workloads often run intermittently, with lower throughput and shorter usage windows. When PTUs... AI - Azure Cognitive Services CER-0510 ### [Regional Standard Azure OpenAI Deployments Without a Data Residency Requirement](https://www.pointfive.co/efficiency-hub/inefficiencies/regional-standard-azure-openai-deployments-without-a-data-residency-requirement) Azure OpenAI models in Microsoft Foundry (formerly Azure AI services) can be deployed as Global, Data Zone or regional (Standard) pay-per-token deployments. The types differ in where prompts and responses are processed: Global may use any... AI - Azure Cognitive Services CER-0511 ### [Missing Commitment Tier for High-Volume Foundry Tools (Azure AI Services)](https://www.pointfive.co/efficiency-hub/inefficiencies/missing-commitment-tier-for-high-volume-foundry-tools-azure-ai-services) Foundry Tools (formerly Azure AI services and Cognitive Services) such as Speech, Translator, Azure Language, Vision OCR and Document Intelligence bill per unit of usage on the default Standard pricing: per audio hour, per million... AI --- Source: the public page above. Product screenshots and illustrative interfaces are examples, not live customer data.