Microsoft announced that its Phi series of generative AI models has a new member - Phi-4. Compared to previous versions, Phi-4 shows improvements in several areas, particularly in solving mathematical problems, thanks to the enhanced quality of training data. As of Thursday evening, access to Phi-4 is relatively limited, available only on Microsoft's newly launched Azure AI Foundry development platform and restricted to research purposes under the Microsoft Research Licensing Agreement. Image source note: Image generated by AI, image licensed.

Microsoft announced the launch of a new tool, Azure AI Foundry, at its annual Ignite conference, aimed at simplifying the operational workflow for cloud customers building and deploying artificial intelligence applications, and enhancing the commercial potential of generative AI. This tool allows users to more easily switch between large language models that support AI. For instance, customers using older OpenAI products can try the new model version or opt for alternatives like Mistral and Meta. Image source note: The image was generated by AI, with image licensing provided by Midj.

Microsoft and GitHub have announced a deep collaboration aimed at simplifying the AI application development process, making it easier for 100 million developers worldwide to build AI applications on the GitHub platform. With the seamless integration of Codespaces and Visual Studio Code on GitHub.com, developers can directly use Azure AI services, including access to leading AI model libraries, direct access to Azure AI through GitHub Models, and harnessing the power of Azure AI.

Microsoft has announced the launch of serverless fine-tuning capabilities for the Phi-3 small language model, enabling developers to easily adjust model performance on the Azure AI platform without managing underlying infrastructure, and it is free for initial use. The Phi-3 model, with 3 billion parameters, is suitable for enterprise developers, offering efficient performance at a lower cost. Although it has fewer parameters compared to OpenAI's GPT-3.5, it performs similarly across various applications. Serverless fine-tuning simplifies operations, reduces barriers, and supports the fine-tuning of small to medium models.