- Microsoft has announced that Deepseek R1 arrives on Copilot + Machines
- It will operate on peripheral, so will be less powerful than the other versions of the model
- Deepseek R1 will have three Copilot + versions that take place over time
Deepseek has seriously shaken the AI world with an LLM which is apparently cheaper to train, more economical in power and yet just as intelligent compared to its competitors. While Meta, Google, Open AI and others rush to decipher how the DEEPSEek R1 model has become so impressive from nowhere – with OpenAi even claiming that he copied Chatgpt to get there – Microsoft takes the “ If you Can’t beat them, join them ”. Rather approach.
Microsoft announced that, after the arrival of Deepseek R1 on Azure Ai Foundry, you will soon be able to run an Optimized NPU version of Deepseek’s IA on your Copilot + PC. This feature will first take place on Qualcomm Snapdragon X machines, followed by Intel Core Ultra 200V laptops and AMD AI chipsets.
It will start by making Deepseek-R1-Distill-qwen-1.5B available on Microsoft AI Takeit for developers, before later unlocking the most powerful 7B and 14B versions. Although these are not as impressive as the 32B and 70B variants also at its disposal, the 14b and lower versions of Deepseek can operate on the devices.
This reduces one of the main concerns concerning Deepseek – that the data shared with AI could be found on unprecedented foreign servers – Microsoft adding that “Deepseek R1 has undergone rigorous red and security team assessments” to reduce more possible security risks.
How to get Deepseek R1 on Copilot +
To start using the Deepseek Copilot + Build, once available, you will need an Azure account – you can register on the official Microsoft website if you don’t already have them. Your next step will be to start the Azure AI foundry and search for Deepseek R1. Then press “Discover the model” on the R1 Deepseek card, before clicking on “deployment” then “deploy” again in the window that appears.
After a few moments, the cat game option should open, and you can start chatting with Deepseek on a device.
If you have not yet used Deepseek, two great advantages that you will find when you install it are that it is currently free (at least for the moment), and that it shows you its `thought ” to measure that he develops his answers. Others, like Chatgpt, go through the same reflection process, but they do not show you, which means that you must refine your prompts thanks to a process of tests and errors until you Get what you want. Because you can see its process and where it could have taken place on the wrong track, you can more easily and precisely modify your guests to achieve your goals.
As the variants 7b and 14b unlock, you should see the Azure model of Deepseek R1 improve, but if you want to test it, you may want to do it sooner than late. Given the serious partnership of Microsoft with Openai, we expect that it does not deal with this emerging rival if it turns out that Deepseek has indeed been copied from chatgpt – potentially delete the Azure, which he may not have the choice of if AI faces a ban in the United States, Italy and other regions.
