Microsoft organizes a team to create small and cheap AI models

Microsoft organizes a team to create small and cheap AI models

Microsoft is working on cheaper and smaller AI models to tap more into this market. The company is already organizing a separate team for such developments.

The new generative AI team will focus on developing small language models (SLMs), which are similar to LLMs such as OpenAI’s GPT-4, but use less computing power. They are planned to be cheaper to operate than OpenAI models.

The new GenAI team will be headed by the corporate vice president of Microsoft, Mykhailo Bilenko, and will become part of the Azure cloud division. The company also transferred several senior AI developers from its research group to it.

Previously, Microsoft presented the Phi-2 model, which, according to the developers, surpasses much larger projects and has about 3 billion parameters. According to experts, the performance of the model is related to the high quality of the data on which it was trained.

Related posts