** **Microsoft has launched the DeepSeek R1 7B and 14B distilled models for Copilot+ PCs via Azure AI Foundry, enhancing AI capabilities at the edge. These models, optimized for Neural Processing Units (NPUs), enable developers to leverage powerful reasoning capabilities directly on their devices, pushing the boundaries of AI innovation.**:**
Unleashing AI Power: DeepSeek R1 Models Now Available
Microsoft has just announced the launch of the DeepSeek R1 7B and 14B distilled models for Copilot+ PCs. This release marks a significant leap in AI capabilities, further integrating artificial intelligence into everyday computing.
What’s New in DeepSeek R1?
The DeepSeek R1 models are designed to run efficiently on Neural Processing Units (NPUs). These models are optimized for local computation, allowing developers to leverage substantial AI power directly from their devices. As Microsoft states,
“Our vision is bold: to build Windows as the ultimate platform for AI innovation.”This vision is becoming a reality with the new models.
Major Updates and Features
These models can handle 7B and 14B parameters, making them capable of complex reasoning tasks. The introduction of NPUs, which can perform over 40 trillion operations per second, enables enhanced performance while minimizing battery drain. This efficiency allows CPUs and GPUs to focus on other tasks, ensuring smooth operation.
The DeepSeek models also benefit from a new scaling law for language models. This law suggests that longer reasoning times can improve response quality significantly. Instead of merely increasing model size or training data, developers can now utilize additional computational power for better outcomes.
What’s Important to Know
Developers can access all variants of the DeepSeek models through the AI Toolkit VS Code extension. The models are available in the ONNX QDQ format, making them easy to deploy locally. Microsoft emphasizes that
“Copilot+ PCs pair efficient compute with the near infinite compute Microsoft has to offer via its Azure services.”This integration allows for seamless transitions between local and cloud-based AI tasks.
Furthermore, the combination of local and cloud capabilities creates a new paradigm of continuous compute. This setup not only enhances performance but also opens doors for innovative applications in various fields.
Get Started Today!
Developers eager to experiment can download the models directly from the AI Toolkit. Simply click the “Download” button and start exploring. The future of AI on Windows looks promising, and Microsoft is excited to see what innovations the developer community will bring.
With the DeepSeek R1 models, the possibilities for AI integration in everyday tasks are limitless. Dive in and unleash the power of AI on your Copilot+ PC!
From the Windows Blog
