Microsoft surprisingly promptly provided its customers for advanced Deepseek R1 AI model with the ability to reflect, writes The Verge. Now it is part of the Azure Ai Foundry and Github models catalog, which allows Microsoft customers to integrate it into their AI applications.
Image source: Solen Feyissa/unsplash.com
«One of the key advantages of using Deepseek R1 or any other model on the Azure Ai Foundry is the speed with which developers can experiment and integrate AI into their work processes, ”said Asha Sharma, Corporate Vice President of AI Microsoft. According to her, Deepseek R1, the announcement of which led to a sharp drop in shares of many US technological companies this week due to the possibility of training with much lower cost compared to the leading models of Openai, passed strict security checks in the Red Teaming format and safety, including automated Assessments of model behavior and extensive safety checks to reduce potential risks.
Microsoft will also soon make a reduced version of R1 available for local launch on the Copilot Plus PC, and it is possible that R1 will appear in other services based on AI from Microsoft.
Earlier there were reports that Openai and Microsoft are investigating about the possible use of the Chinese company API Openai to teach its models.