--- id: "2026-11-18-microsoft-aion-1-0-windows-slm-open-weights" url: "https://postcutoff.com/e/2026-11-18-microsoft-aion-1-0-windows-slm-open-weights/" as_of: "2026-11-22T13:32:00+01:00" date: "2026-11-18" date_precision: day category: open-source importance: 2 confidence: high status: [Confirmed] sources: 2 editor: human editor human_review: null version: "2026-11-19" --- As of: 2026-11-22 13:32 CET. Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/ Canonical page: https://postcutoff.com/e/2026-11-18-microsoft-aion-1-0-windows-slm-open-weights/ # Microsoft publishes Aion 1.0 Instruct (2.3B, MIT licence), the small model that is to become Windows' built-in on-device LLM by end of January 2027 On 18 November 2026 Microsoft's Windows Applied Sciences Group put Aion 1.0 Instruct, a 2.3B-parameter small language model, on Hugging Face under the MIT licence. Microsoft says a further optimised version, paired with its content-moderation model, will become the Windows OS small language model by the end of January 2027; the release is an early look. ## Key facts - Hugging Face: microsoft/Aion-1.0-Instruct-2.3B, created Nov 18, 2026, ~2.29B parameters (safetensors), MIT licence; unquantized weights first, quantized versions to follow; PyTorch, ONNX and GGUF listed - 32K-token context; eight languages (English, German, Spanish, Portuguese, Italian, French, Chinese, Japanese); aimed at on-device summarisation, rewriting, chat, document comprehension and OS tasks - Model card: 'A further optimized version of Aion Instruct, coupled with Microsoft's content moderation model, will become the Windows OS SLM by the end of January 2027'; sample repo microsoft/Aion-Instruct-Preview-Sample - Found by the sweep's new Hugging Face lab-organisation check (54 likes, 788 downloads on Nov 19); no press coverage found yet ## What happened Microsoft quietly released the weights of the small model it plans to ship inside Windows, two months before that rollout. ## Why it matters An operating-system-level LLM on hundreds of millions of PCs would be one of the most widely deployed models; publishing its weights openly lets outsiders test it first. ## Your AI and this story - GPT-6.2 Astra (training cutoff May 2026): 171 days after its cutoff - Claude Opus 5.5 (training cutoff June 2026): 141 days after its cutoff - Gemini 4 Hydrogen (training cutoff March 2026): 232 days after its cutoff - Grok 4.7 (training cutoff May 2026): 171 days after its cutoff ## Sources 1. [Hugging Face: microsoft/Aion-1.0-Instruct-2.3B](https://huggingface.co/microsoft/Aion-1.0-Instruct-2.3B) (huggingface.co, code) 2. [GitHub: microsoft/Aion-Instruct-Preview-Sample](https://github.com/microsoft/Aion-Instruct-Preview-Sample/) (github.com, code) ## Changes - 2026-11-19 (filed): Created from the Hugging Face model card ## Related - 2026-11-17: [Ignite 2026: MAI Code 2, Office in Copilot](https://postcutoff.com/e/2026-11-17-microsoft-ignite-mai-code-2-copilot-office/index.md)