Microsoft Enhances Copilot+ PCs with Distilled DeepSeek R1 Models

Microsoft has announced the availability of DeepSeek R1 7B and 14B distilled models for Copilot+ PCs via Azure AI Foundry. These optimized AI models offer similar intelligence to larger models but require significantly less computing power, making AI-powered features faster and more efficient on standard hardware.

What Are Distilled Models and Why Do They Matter?

Previously, Microsoft introduced DeepSeek R1, a powerful AI model capable of handling complex tasks on Copilot+ PCs. However, running such a large model requires significant computing power, making it difficult for everyday devices to operate smoothly.

To address this, Microsoft now offers distilled versions of DeepSeek R1. These models retain the knowledge of the original but are optimized for faster, more efficient performance.

Think of it like a teacher-student relationship:

• The original DeepSeek R1 model (the “teacher”) trains smaller versions (the “students”).
• These students learn the same concepts but execute tasks more efficiently on specific hardware.

This allows AI features to run directly on a PC without relying on cloud computing, ensuring faster responses, better privacy, and lower power consumption.

How Will This Impact Users?

For Developers:

• AI models can run locally on a PC, ensuring real-time responses without needing an internet connection.
• Developers can build faster, smarter applications, including virtual assistants, speech recognition tools, and automation systems that work instantly.

For Everyday Users:

• AI-powered tools like email drafting, document summarization, and scheduling assistants will run faster and more efficiently.
• Running AI models on-device ensures longer battery life and improved privacy, keeping sensitive data from cloud servers.

How Copilot+ PCs Run AI Models Locally

The Neural Processing Unit (NPU) is the key technology enabling on-device AI processing. Unlike traditional CPUs (for sequential processing) and GPUs (for graphics workloads), NPUs are designed specifically for AI tasks, making them:

• Faster: Handle AI workloads with minimal latency.
• Power-Efficient: Reduce battery drain and prevent overheating.
• Optimized: Free up CPUs and GPUs for other tasks, improving overall system efficiency.

Which Copilot+ PCs Will Support Distilled Models?

The DeepSeek R1 distilled models will first launch on Copilot+ PCs powered by Qualcomm Snapdragon X. Support will later expand to devices featuring Intel Core Ultra 200V and AMD Ryzen processors.

With these advancements, Microsoft is making AI more accessible, efficient, and powerful on everyday computing devices.