Microsoft is expanding its AI capabilities by incorporating the DeepSeek-R1 AI model into Windows Copilot+ PCs. This move complements Microsoft’s existing partnership with OpenAI’s ChatGPT and demonstrates the company’s commitment to exploring diverse AI solutions.
The DeepSeek-R1 model will be optimized for Snapdragon X devices initially, with support for Intel Lunar Lake and AMD Ryzen AI 9 PCs to follow. Developers will also gain access to the DeepSeek-R1-Distill-Qwen-1.5B model through the Microsoft AI Toolkit, with more powerful 7B and 14B variants expected later.
Microsoft emphasizes the NPU-optimized nature of these models, enabling efficient on-device AI applications that leverage the capabilities of Copilot+ PCs. The company has implemented low-bit processing to ensure smooth local execution on NPU hardware.
In addition to Windows Copilot+ PCs, DeepSeek’s R1 model will be available on Microsoft’s Azure AI Foundry platform alongside other leading AI models like GPT-4, Mistral AI, and Meta-Llama 3.
Microsoft acknowledges privacy concerns surrounding the China-based DeepSeek and assures users that the R1 model has undergone rigorous safety evaluations and security reviews to mitigate potential risks.
DeepSeek’s R1 model has disrupted the AI industry with its efficient use of computing power. This has led to market fluctuations and concerns from OpenAI regarding the alleged “distillation” of its models.