Microsoft Azure AI Foundry

Microsoft Foundry is Azure's platform for deploying models, including OpenAI models sold by Azure, and for building agents on them. Microsoft's Foundry pricing page says the platform is free to use and explore, but each service inside it has its own billing model and price. We could not list per-token prices: the model pricing page we opened on 5 November 2026 showed placeholders instead of figures, and it says prices are estimates only. It does list deployment types, including Global, Data Zone and Regional, plus provisioned throughput, where GPT-5.5 needs at least 15 provisioned throughput units on Global or Data Zone and 50 on Regional. A one-year agent pre-purchase plan gives 5, 10 or 15 percent off at 20,000, 100,000 and 500,000 units. On data, Microsoft says prompts, completions and embeddings are not available to other customers or to OpenAI, and are not used to train foundation models without your permission. When the abuse system sees warning signs, a sample of prompts may go to human review unless you are approved for modified monitoring. Global deployments can process data in any geography, and Data Zone deployments within one zone. Facts come from Microsoft's pricing pages and data privacy page we checked on 5 November 2026.