Xiaomi Open-Sources Robotics-U0 Embodied World Model and Training Stack

Xiaomi released Xiaomi-Robotics-U0, an autoregressive embodied model with ~38B parameters trained on image and video generation. It published U0-4B and FlashAR weights, inference code, a Gradio demo, and FSDP training releases; FlashAR+ speeds 1024×1024 generation by ~83x.

• Main line ~38B parameters; open weights U0-4B and FlashAR
• FlashAR+ speeds 1024×1024 generation from 450.8s to 5.44s (~83x)
• Model combines text-to-image, editing, multi-view scenes, embodied video
• Interference task success rose from 36.9% to 63.2% on new backgrounds and light