Microsoft Aion Instruct: Local AI on Windows Expands to GPU

Today we’re sharing the transition plan from Phi Silica to Aion Instruct, Microsoft’s latest on-device language model for Windows. Aion Instruct will begin rolling out to CP+ devices through the Windows Insider Program in late November, with full Windows Retail availability targeted for late January. We’re also expanding Aion Instruct to supported GPUs, bringing local language model capabilities to a broader range of Windows devices.
Microsoft Aion Instruct is designed to power fast, responsive language model experiences directly on your device. This transition gives Windows and Microsoft applications a newer, more efficient model foundation while preserving the familiar experiences users rely on today.
You can also tune in next Tuesday to THR888 at Ignite 2026 to learn more about Aion Instruct (recording will be available).
Local AI enables applications to deliver responsive experiences while keeping processing on device. Aion Instruct builds on our investment in this area with improvements designed to help language model features run efficiently alongside the applications and workflows you already use on Windows.
A technical report with more details on Aion Instruct is coming soon to the Microsoft Command Line blog.
As Aion Instruct becomes the foundation for more Windows experiences, users can expect:
For most app developers, no action will be required. The change will happen behind the scenes as supported experiences move to Aion Instruct. Action will be required for apps that rely on highly custom prompting (ex. Specific system prompts) and custom LoRAs, as well as apps that have custom model output parsing due to our new watermarking implementation. We’ve added invisible watermarking to Aion Instruct to fulfill Text Provenance requirements in specific regions. Additionally, apps that are heavily reliant on Phi Silica’s specific behavior or output strings should be validated for quality. Since Aion Instruct is a new model, output strings and model behavior are different. To get the most out of Aion Instruct, take a look at our prompt guidance.
Like Phi Silica today, we’re also expanding Aion Instruct to supported GPUs, including popular dGPUs like NVIDIA’s RTX-series and AMD’s RDNA4 generation. This greatly increases the number of Windows devices that can benefit from Aion Instruct and gives developers greater flexibility when building language model experiences that run on-device.
Our goal is to provide a consistent local AI foundation across supported Windows hardware. In many scenarios, applications can take advantage of available device capabilities without developers requiring separate end user experiences for each type of hardware.
Hardware support will expand in stages as configurations meet our readiness and quality standards.
For developers building local language model experiences on Windows, Microsoft Aion Instruct provides a newer foundation while continuing to work with familiar Windows AI capabilities and APIs.
We encourage developers to begin evaluating their applications with Aion Instruct as it becomes available. In particular, developers should validate model-dependent behavior, review any application-specific customizations, and confirm that key scenarios continue to deliver the intended experience. Aion Instruct can be validated end to end now with the package available at aka.ms/tryaion. Additionally, the Windows Insider Program Beta External build targeted for release on 11/20 will allow full, end to end validation.
Because language models can differ in how they interpret and produce content, testing is an important part of any model transition–even when the underlying APIs remain familiar.
We’ll provide additional documentation and migration guidance as availability expands.
The transition will roll out in stages:
Developers can find the latest transition information in the Windows AI documentation:
https://learn.microsoft.com/en-us/windows/ai/apis/phi-silica
Phi Silica helped establish a foundation for local language-model experiences on Windows. Microsoft Aion Instruct represents the next step in that journey, improving the platform while enabling it to reach more devices. With Aion Instruct, we’ll also be removing key adoption barriers like the need for an LAF token (removal targeted for January).
The staged transition gives developers time to evaluate their experiences and prepare for broader adoption. Most existing AI-powered features will continue to work (with the exception of custom LoRAs) as they move to the newer model foundation.
This is the beginning of the next phase of local AI on Windows. We’ll continue expanding Aion Instruct across supported hardware and working with developers to improve the experience.
As the rollout progresses, we’ll share updated guidance, documentation, and ways for developers to try Aion Instruct and provide feedback.
We’re excited to see what you build with Aion Instruct on Windows. Be sure to check out THR888: Build On-Device AI Experiences w/ Windows AI APIs at Ignite 2026 to learn more about Aion Instruct (recording will be available).
To share feedback, create an Issue in the WinAppSDK repo: Issues · microsoft/WindowsAppSDK
AI Platform and Tools team.
Be the first to start the discussion.
Enter the destination URL
Or link to existing content