Hy-Embodied RxBrain-1.0

Tencent's embodied/robotics foundation model (~6.2B): a Mixture-of-Transformers with modality-specific pathways and a flow-matching image head over a FLUX VAE latent, doing embodied understanding, world-state prediction, and subgoal planning (both action and goal-image) for robotic manipulation and spatial reasoning. Ships with a Hy-Embodied-VLM-1.0 sibling — an on-thesis domain foundation model.