DeepSeek to be Supported by Windows Copilot+ PCs

Microsoft has announced it is bringing the DeepSeek-R1 AI model to Copilot+ PCs in the near future. While maintaining its current relationship with OpenAI’s ChatGPT model, the company will offer DeepSeek-R1 on Snapdragon X devices and later Intel Lunar Lake and AMD Ryzen AI 9 PCs.

In a blog post, Microsoft reveals that the DeepSeek-R1-Distill-Qwen-1.5B model will be supported on Copilot+ PC devices “soon”. The company went on to state that it’ll be available on Microsoft AI Toolkit for developers. 7B and 14B variants will be arriving soon after the initial release.

Microsoft Copilot+ PC is equipped with at least 256GB of storage, 16GB of RAM and an NPU supporting 40 TOPS of power. “The optimized DeepSeek models for the NPU take advantage of several of the key learnings and techniques from that effort, including how we separate out the various parts of the model to drive the best tradeoffs between performance and efficiency, low bit rate quantization and mapping transformers to the NPU,” the blog post states. The 1.5B model of DeepSeek isn’t as powerful compared to the higher 32B and 70B models. However, Microsoft makes it sound as if they are”NPU-optimized” for Copilot+ PCs.

DeepSeek’s R1 model is also being brought aboard Microsoft’s Azure AI Foundry platform. Earlier reported by The Verge, the company is moving swiftly to onboard the model alongside GPT-4, Mistral AI, and Meta-Llama 3. With these integrations, Microsoft’s customers could integrate the R1 model into their AI applications.

What’s notably interesting about Microsoft’s willingness to work with DeepSeek is that it’ll support the R1 model in various ways at the same time it’s reportedly probing the company. Microsoft and OpenAI are investigating whether DeepSeek has obtained data from OpenAI’s technology through unauthorized means.

This week, DeepSeek overtook ChatGPT as the number one free app in Apple’s App Store.