Xiaomi, a major Chinese technology company, has recently introduced two new artificial intelligence (AI) models called MiMo-V2.6 Pro and MiMo-V2.6-Flash. These models are embedded in Xiaomi’s smartphones and are designed to process various types of data, including text, images, video, and even computer code. The MiMo-V2.6 Pro is the more powerful of the two, while the Flash version focuses on delivering faster responses and reducing the cost of using the AI. One of the standout features of these models is their ability to understand and generate three-dimensional (3D) environments. This is made possible through a function called "Vibe World," which allows the system to create interactive 3D spaces based on text, images, or video input. This technology can be used to develop video games, design 3D models for software like Blender, compose music in demo form, or even control robotic arms using input from multiple cameras. According to Xiaomi, the MiMo-V2.6 Pro model ranks first among open-weight AI models on the Artificial Analysis Intelligence Index benchmark. It outperforms models like Alibaba’s Qwen 3.8 Max but is slightly behind other high-end models such as Claude Opus 5, GPT-5.6 Sol, and Meta Muse Spark. These comparisons are based on the maximum performance of each model. Xiaomi has made the model weights, over 7,000 reinforcement learning (RL) environments, and its full RL infrastructure available as open-source. This means developers and researchers can access and use these resources freely. The MiMo-V2.6 models are available on popular AI platforms such as HuggingFace and OpenRouter, and can also be accessed via API at the same rate as the earlier V2.5 Pro version. In addition to mobile integration, Xiaomi is also expanding its MiMo desktop applications to support Windows and macOS operating systems, broadening the accessibility of these AI tools for both mobile and desktop users.