The ALLEX simulation model is designed to bridge the persistent gap between virtual environments and physical performance. By precisely replicating the robot’s high backdrivability and force transparency, the model allows developers to conduct sophisticated motion learning that translates directly to real-world deployment. The release supports industry-standard formats, including MJCF for MuJoCo, USD for Isaac Sim, and URDF for ROS, ensuring compatibility across major robotics development platforms.

Co-CEO Yong-Jae Kim emphasized that creating a robust simulation infrastructure is essential for scaling humanoid capabilities. The company intends to follow this initial release with a broader suite of core technologies from its humanoid program throughout the year, culminating in the launch of the physical ALLEX research platform. This strategy follows a period of rapid growth for the firm, which secured KRW 95 billion in Series B funding earlier in 2026 and earned recognition through the NVIDIA and AWS-led Physical AI Fellowship.