Microsoft's 2.3B Aion Instruct hints every Windows device will ship with an on-device LLM by 2027
NielsRogge · x · 2026-11-19
Microsoft released Aion Instruct on Hugging Face: a 2.3B-parameter small language model built for on-device AI in Windows, with a 32K context window and support for 8 languages.
The model card states that "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." Hugging Face's NielsRogge reads this as Microsoft planning to ship an LLM on every Windows device.
Related event: Microsoft releases Aion-1.0-Instruct-2.3B for on-device Windows AI(3 posts)→
More from Models
- Code Arena launches WebDev leaderboard as GPT-6.2 Astra takes #1 in Consumer Product — arena · 2026-11-19
- Code Arena leaderboard: GPT-6.2 Astra climbs to #1 in Consumer Products over GPT-6 Astra — arena · 2026-11-19
- GPT-6.2 Astra and Sol join Code Arena Pareto frontier: +40 pts over Claude at 53% lower cost — arena · 2026-11-19
- Leaker hints Gemini 3.8 Flash already in Stitch, next-gen "Carbon 4" will be much faster — testingcatalog · 2026-11-19
- OpenAI dev account showcases GPT-6.2 ASTRA building an app in 30 seconds from one prompt — OpenAIDevs · 2026-11-19
- A 'SOTA' LLM missed 4 glaring grammar errors after 10 proofreading passes on a 3,000-word article — Lokoto123 · 2026-11-19