Microsoft Azure
Microsoft Azure is Microsoft's cloud platform. For LLMs it offers open and partner models through Azure AI Foundry, managed endpoints in Azure Machine Learning, GPU virtual machines, and managed Kubernetes (AKS) for running models at scale.
Overview
Microsoft Azure is Microsoft's public cloud, with data centres in more than 60 regions across the Americas, Europe, the Middle East, Africa and Asia Pacific.
It gives you four main ways to run large language models. Azure AI Foundry offers ready-to-use models, including open-weight models like Llama, Qwen and gpt-oss, billed per token or with reserved throughput. Azure Machine Learning deploys open or fine-tuned models to managed endpoints. Azure GPU virtual machines, including NVIDIA H100, H200 and GB200 and AMD MI300X, let you run any model yourself, and Azure Kubernetes Service (AKS) runs and scales model servers on those GPUs.
Azure suits businesses already on Microsoft 365 or Azure, those that need enterprise security and compliance, and teams that want OpenAI and open models in one place.
Plans from Microsoft Azure
| Hosting plan | Hosting type | Pricing range | Model families |
|---|---|---|---|
| Azure AI Foundry | Serverless API | US$0.15 – US$1 | Llama, Qwen, gpt-oss |
| Azure Machine Learning | Dedicated endpoint | From US$0.526 | gpt-oss, Qwen, DeepSeek, Llama, Gemma, Mistral |
| Azure GPU Virtual Machines | GPU cloud | From US$0.526 | gpt-oss, Qwen, DeepSeek, Llama, Gemma, Mistral |
| Azure Kubernetes Service (AKS) | GPU cloud | From US$0.526 | gpt-oss, Qwen, DeepSeek, Llama, Gemma, Mistral |