Microsoft Brings DeepSeek R1 to Windows 11 Copilot+ PCs

Microsoft is bringing NPU-optimized variations of DeepSeek R1 to Copilot+ PCs.
– DeepSeek R1 is first coming to Qualcomm Snapdragon X-powered devices, followed by Intel and AMD AI chipsets.
– The DeepSeek-R1-Distill-Qwen-1.5 B design is being contributed to the Microsoft AI Toolkit.

Microsoft has revealed that it’s bringing NPU-optimized variations of DeepSeek R1 to Copilot+ PCs. The business will likewise integrate the DeepSeek-R1-Distill-Qwen-1.5 B design into its Microsoft AI Toolkit for designers, with the 7B and 14B versions set to follow.

In a recent article, Microsoft revealed that DeepSeek R1 designs will first be readily available on Qualcomm Snapdragon X-powered PCs, followed by Intel Core Ultra 200V laptop computers and AMD AI chipsets. This release will allow developers to develop AI-powered apps that run in your area on compatible Copilot+ PCs.

„The optimized DeepSeek designs for the NPU take advantage of numerous of the key knowings and methods from that effort, including how we separate out the different parts of the design to drive the very best tradeoffs in between performance and efficiency, low bit rate quantization and mapping transformers to the NPU,” Microsoft described.

Microsoft has actually laid out the hardware requirements for running these AI models on Windows 11 devices. To certify, a PC needs to have a Neural Processing Unit (NPU) with at least 40 TOPS (trillion operations per second), 16GB of RAM, and 256GB of storage. This implies that PCs with old NPUs won’t have the ability to run these models locally.

How to run DeepSeek R1 on Windows 11 Copilot+ PCs?

To begin with DeepSeek on a Copilot+ PC, developers will require to an Azure account on Microsoft’s site. Now, launch Azure AI Foundry and then look for DeepSeek R1. Select the „Have a look at design” choice, click Deploy, and then click „Deploy” once again in the pop-up window. The Chat Playground option will appear, and developers can begin try out DeepSeek R1 in your area on their Copilot+ PCs.

Microsoft has likewise revealed that it’s making the open-source DeepSeek R1 LLM readily available for developers through Azure AI Foundry and GitHub. „Among the crucial advantages of using DeepSeek R1 or any other design on Azure AI Foundry is the speed at which designers can experiment, iterate, and incorporate AI into their workflows. With built-in design evaluation tools, they can quickly compare outputs, benchmark performance, and scale AI-powered applications,” stated Asha Sharma, Corporate Vice President, AI Platform.

A brand-new report from the Financial Times reveals that Microsoft is investigating whether Chinese start-up DeepSeek illegally used OpenAI’s information to train its R1 model. This action violates OpenAI’s regards to service, and Microsoft prepares to work together with the US federal government to secure its AI model.

Microsoft’s statement aims to attend to concerns about DeepSeek potentially storing data on unsecured foreign networks. To reduce this risk, the company has subjected DeepSeek R1 to extensive red teaming and safety evaluations to lower the threat of data breaches.