Azure Machine Learning

Azure Machine Learning is Microsoft's cloud service for building, training and deploying machine learning models. I did not reach an official Microsoft pricing page, so this relies on third-party guides that disagree in places. The common thread is that you pay for the underlying compute and services rather than for the workspace itself, and that compute is billed separately from tokens. One source says Azure ML compute instances keep billing until explicitly stopped, and dedicated endpoints run continuously so you pay for their instances around the clock. Another estimates that Azure ML prices GPU clusters at about 25% above raw Azure VM on-demand rates, a figure I could not corroborate.

There is a naming change worth knowing: Azure AI Foundry became Microsoft Foundry after January 1, 2026. One guide says Foundry itself is free with you paying for the services it orchestrates, while another says pricing is sales-led with no self-serve tiers published. Model inference there is mainly token-based, with one source listing GPT-4o at about $2.50 per million input and $10 per million output tokens, and provisioned throughput units offered for high-volume work. Supporting services such as AI Search and networking add separate charges, so a claim of no additional charge does not hold for compute. For RAG-only deployments one guide says Azure ML compute cost is minimal.

Azure Machine Learning suits organisations already on Microsoft Azure. Compare Amazon SageMaker, Vertex AI and Kubeflow. Browse the MLOps & Infrastructure tools directory for more.

ML on Azure

Train and deploy.

Foundry-based AI apps

Models and RAG.

Is Azure ML free?

Compute and endpoints are billed.

Is Foundry renamed?

Yes, Microsoft Foundry since January 2026.

Do instances stop automatically?

No, stop them explicitly.

Token prices?

e.g. GPT-4o about $2.50/$10 per million.