Microsoft is stepping into the spotlight with its latest announcement, embracing DeepSeek’s controversial R1 AI model despite swirling investigations. The tech giant is set to introduce an NPU-optimized version of this model to its lineup of Copilot+ PCs. The first release, known as DeepSeek-R1-Distill-Qwen-1.5B, has been integrated into Microsoft’s AI Toolkit for developers, signaling a cautious yet progressive approach towards this innovative technology.
DeepSeek has garnered significant attention in the tech community, and Microsoft is keen on harnessing its potential. Initially, the R1 model will be tailored for devices using Qualcomm’s Snapdragon X chipsets, followed by compatibility with Intel’s Core Ultra 200V and other platforms. As excitement builds, the tech world eagerly anticipates the upcoming rollouts of the more advanced 7B and 14B models, promising developers an expansive array of tools for exploration and development.
For those eager to dive into DeepSeek’s capabilities on a Windows 11 system, getting started is straightforward. Microsoft offers simple guidance:
1. Download the AI Toolkit VS Code extension.
2. Access the DeepSeek model, optimized in the ONNX QDQ format, available soon in the AI Toolkit’s model catalog.
3. Download the model directly from Azure AI Foundry by selecting the “Download” option.
4. Once downloaded, users can experiment with the AI by launching the Playground, loading the “deepseek_r1_1_5” model, and sending it prompts.
Microsoft has invested in multiple optimizations to ensure DeepSeek’s R1 operates smoothly and effectively on NPU-based hardware. While the ongoing investigation into whether the R1 model was trained using OpenAI data continues, Microsoft’s development and integration efforts proceed unhindered. We’ll keep a close eye on any new developments in this story and provide updates as they emerge. Stay connected for the latest news in this evolving AI saga.
