Microsoft's July 2026 account of Hugging Face models on Foundry describes moving deployment preparation into a managed service. For its curated collection, Microsoft stages model weights in Azure and builds and scans the software containers that serve them. Developers choose a deployment template specifying the runtime, hardware and context windowThe maximum amount of text a model can consider at once — its working memory for the current conversation or task.Full definition → settings. The article says this preparation allows deployment without outbound access to Hugging Face Hub, a useful property for private networks. The collection was described as a preview, with support for bringing custom weights still on the roadmap.
Terms in this piece · Glossary
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
