On Wednesday, Microsoft announced the integration of the newly-launched DeepSeek-R1 artificial intelligence (AI) model into its Azure AI Foundry and GitHub platforms. This reasoning-driven AI model is now available for access, marking a significant addition for users of these services. The technology firm, based in Redmond, has reassured users by stating that it has conducted thorough safety assessments to confirm the model’s suitability for both personal and commercial applications. Furthermore, distilled versions of the R1 model will soon be accessible to Copilot+ PC users through the AI Toolkit.
Microsoft Introduces DeepSeek-R1 in Azure AI Foundry and GitHub
In a blog post, Microsoft revealed that the DeepSeek-R1 AI model has been added to the model catalogue available on Azure AI Foundry and GitHub. Azure AI Foundry is tailored for enterprises, enabling developers to create, test, and develop generative AI applications and customized copilots.
Asha Sharma, Microsoft’s Corporate Vice President of AI Platform, emphasized the benefits of using the DeepSeek R1 model within Azure AI Foundry, stating, “One of the key advantages of using DeepSeek R1 or any other model on Azure AI Foundry is the speed at which developers can experiment, iterate, and integrate AI into their workflows.”
The announcement also highlighted the extensive red teaming and safety evaluations the R1 model underwent, which included automated assessments of its behavior and comprehensive security reviews. Microsoft reassured users that it has taken measures to limit potential risks. Additionally, Azure AI Foundry enhances safety with its Content Safety filtering system and the Safety Evaluation System.
To access DeepSeek-R1, users can search for it within the model catalogue, open the model card, and click on deploy. This action will provide the necessary inference application programming interface (API) to utilize the model.
Distilled Versions of DeepSeek-R1 Arriving for Copilot+ PC Users
In a separate post, Microsoft announced that optimized versions of DeepSeek-R1 are being introduced for Copilot+ PC users. Initially, these distilled models will be available on devices powered by Snapdragon X, followed by support for Intel Core Ultra 200V chipsets and others. The first release will feature the DeepSeek-R1-Distill-Qwen-1.5B model, accessible through the AI Toolkit, with the 7B and 14B variants slated for later inclusion.
