In early 2025, Microsoft plans to introduce Phi Silica, a Small Language Model (SLM), to Windows PCs as part of its Copilot feature.
This move, announced at CES 2025 in Las Vegas, will allow users to run AI-powered applications directly on their PCs, rather than relying on cloud-based models.
Phi Silica, which debuted at Microsoft Build 2024, is a 3.3 billion parameter model designed to run locally on PCs.
Unlike Large Language Models (LLMs), which are typically cloud-dependent and require expensive subscriptions, SLMs like Phi Silica provide a more private, cost-effective solution for AI chatbots and other applications.
With this development, Windows Recall and other AI features will be able to operate using local AI capabilities, ensuring privacy and preventing data from being transmitted to the cloud.
The integration of Phi Silica is expected to enhance the speed and accuracy of AI-driven tools on Windows.