## Xiaomi Unveils Open-Source Robotics Model, “Robotics-0,” Bridging understanding and Action
Xiaomi has released an open-source Visual Language action (VLA) model, dubbed “Robotics-0,” designed to improve the ability of robots to understand human instructions and translate them into coordinated physical actions. [https://www.gizmochina.com/2026/02/12/xiaomi-robotics-0-open-source-vla-model/](https://www.gizmochina.com/2026/02/12/xiaomi-robotics-0-open-source-vla-model/)
The model operates in two key stages: first, it processes human instructions and perceives the surrounding environment through visual input. Second, it converts this understanding into precise and coordinated movements. A key feature highlighted by Xiaomi is the model’s ability to maintain contextual understanding even while executing physical tasks – a challenge for many existing systems that can lose cognitive function when focused on movement.Xiaomi claims the Robotics-0 model delivers smooth, uninterrupted responses. In simulated environments like LIBERO, CALVIN, and SimplerEnv, it reportedly outperformed approximately 30 other models. [https://www.gizmochina.com/2026/02/12/xiaomi-robotics-0-open-source-vla-model/](https://www.gizmochina.com/2026/02/12/xiaomi-robotics-0-open-source-vla-model/)
The model has also been tested on a physical, dual-arm robot, demonstrating its ability to reliably perform tasks such as folding towels and manipulating blocks with precision.
by releasing Robotics-0 as an open-source project, Xiaomi aims to foster collaboration within the research community and accelerate the development of more adaptable robots. The move raises the question of how quickly increasingly refined robots will become integrated into everyday life as more technology companies invest in this field.