Microsoft introduces DeepSeek R1 model into Copilot+PC: AI intelligent new situation

Recently, Microsoft announced that it will introduce the optimized DeepSeek R1 model in its Copilot+ personal computers. This decision not only marks a crucial step for DeepSeek to expand from mobile devices to the Windows system but also represents an important part of the software giant's new round of layout in the field of artificial intelligence. This new technology is expected to be widely applied in devices equipped with Snapdragon X chips and Intel Core Ultra 200V processors, and in the future, it will also support personal computers based on AMD Ryzen AI9.

The DeepSeek R1 model has been refined, with the first product being DeepSeek-R1-Distill-Qwen-1.5 billion. Compared to its full model with 671 billion parameters, although its capabilities are somewhat reduced, its more compact design enables the model to run efficiently on consumer-grade hardware. It should be noted that this model refinement, or 'knowledge distillation' process, aims to extract key knowledge from large models for use in smaller models, thereby reducing hardware requirements and enhancing user experience.

Microsoft emphasized the adaptability of the optimization process to different hardware architectures, highlighting the overall performance improvement. This model is particularly suitable for running in environments with neural network processors (NPU), with the response and processing speed significantly enhanced, reducing the time to generate the first token to 130 milliseconds and enabling it to process up to 16 tokens per second, further improving user interaction efficiency.

The introduction of this technology will not only enhance the AI experience on personal computers but also drive the application of AI in daily office work, creation, and other fields. For example, when users utilize Copilot+ PC, they can use the DeepSeek R1 model for text generation or image creation, achieving higher accuracy and faster speed. This provides strong support for writing, design, and other creative work, further expanding productivity.

Microsoft's support and layout in the AI field have also attracted much attention. As an important investor in OpenAI, Microsoft offers a variety of options on its Azure Playground platform, including GPT models, Meta's Llama models, and the newly introduced DeepSeek model, demonstrating its comprehensive promotion of AI technology openness. This diverse strategy not only helps developers make choices based on different needs but also promotes the development of more intelligent applications and services.

In addition, for users who plan to run AI locally, Microsoft also provides an AI toolkit for VSCode, from which users can download various DeepSeek models, making it simple and convenient. For instance, with 'deepseek_r1_1_5', users can easily obtain and implement the files, allowing ordinary users to enjoy the power of AI with low barriers.

As AI technology continues to develop, fields such as AI painting and writing are also receiving increasing attention. Compared to traditional creation methods, AI tools have significantly improved creation efficiency. By making reasonable use of the newly introduced DeepSeek R1 model by Microsoft, users can generate content and design images more efficiently, driving innovation in various industries. However, as AI technology advances, how to effectively manage the changes and challenges brought by AI and ensure its safe and beneficial use is a question that every technology worker needs to deeply consider.

Overall, by introducing the DeepSeek R1 model, Microsoft not only enhances the intelligence level of Copilot+ PC but also lays the foundation for the wide application of AI. In the future, we look forward to seeing the positive changes this technology brings to more fields and how it further boosts personal productivity. For more information, you can visit Microsoft's official website or directly explore various AI tools and models on the Azure Playground platform.