Mu by Microsoft is a new ~330M on-device SLM that runs on the NPU in Copilot+ PCs. It powers the new agent in Windows Settings, delivering high speed and efficiency for local AI tasks.
Hi everyone!
On-device AI is developing at an incredible pace lately. We keep seeing models that are smaller, yet more powerful, and specifically optimized for edge devices and chips. This is great news, especially for developers building privacy-focused applications or new AI hardware.
Microsoft's new model, Mu, is built for exactly this kind of on-device work. It's a small language model that runs locally on the NPU in Copilot+ PCs, and it's incredibly fast, powering the new agent in Windows Settings.
What’s really exciting here is that this is just the first step. Mu gets its capabilities through distillation from the larger Phi models. This signals a future where on-device AI can keep getting smarter without a huge increase in size.
It's seriously making me think it might be a very good time to get a Copilot+ PC.👀
Exciting times ahead with the new Microsoft Copilot and Mu! It’s amazing to see on-device AI becoming more capable and privacy-focused. The idea of running such a powerful model locally without compromising speed is impressive. How do you think this will impact the development of personalized AI apps in the future, especially in terms of performance and privacy?
Replies
Exciting times ahead with the new Microsoft Copilot and Mu! It’s amazing to see on-device AI becoming more capable and privacy-focused. The idea of running such a powerful model locally without compromising speed is impressive. How do you think this will impact the development of personalized AI apps in the future, especially in terms of performance and privacy?