Beijing-based multimodal AI company, founded 2023, that builds the HiDream-O1 model family (Image, World, Embodied) on its self-developed UiT (Unified Transformer) architecture, moving from simulated-world modeling toward real-world embodied deployment.
YA
Yao Ting (姚霆), Co-founder & CTO of HiDream.ai
Public face of the launch, framing HiDream-O1-Embodied as the milestone that moves the company's world-model strategy from 'simulated world' to 'real world'; previously led billion-scale image search and logistics robotic-arm vision systems at JD.com.
ME
Mei Tao (梅涛), Founder/CEO of HiDream.ai
Sets the strategy positioning world models as the 'cognitive hub' linking data, world models, agents, and embodied execution, arguing embodied AI's data flywheel cannot spin without a high-quality world model.
Motion-capture and physical-validation partner supplying high-precision mocap data that HiDream.ai's generative video technology expands roughly 100x while preserving physics constraints, feeding embodied-AI training data.
Third-party embodied-AI evaluation platform, open to universities and research institutions worldwide, whose disturbance-adaptation sub-leaderboard ranking gives HiDream-O1-Embodied external, comparative validation.