Microsoft Open Sources Multi-Modal AI Agent - Magma

At 3 a.m. Singapore time, Microsoft open-sourced the multimodal AI Agent base model - Magma on its official website. Compared with traditional Agents, Magma has multimodal capabilities across digital and physical worlds, able to automatically process different types of data such as images, videos, texts etc. For example, you can use Magma to automatically place e-commerce orders or check the weather; it can also operate physical robots automatically or provide assistance when playing real chess games. In addition, Magma also has built-in psychological prediction functions that enhance its understanding of spatiotemporal dynamics in future video frames and accurately predict the intentions and future behaviors of characters or objects in videos.