BeingBeyond, a leading developer of general-purpose embodied foundation models, recently announced a strategic partnership with Daimon Robotics.
The partnership brings together the two companies' complementary strengths in models, perception, and data. By combining Daimon Robotics' full-stack Physical AI capabilities — including its tactile-grounded world model, VTLA multimodal data, and multidimensional vision-tactile perception — with BeingBeyond's leading algorithms for general-purpose embodied foundation models, the two companies will collaborate on multimodal data development, model training and evaluation, and benchmark scenario validation. The partnership aims to accelerate the large-scale deployment of robotics in areas such as 3C precision manufacturing and flexible production.
BeingBeyond focuses on the research, development, and application of general-purpose embodied foundation models. It was among the first companies in China to propose a general-purpose embodied model framework trained on large-scale human video data, and is also the first AI startup in China to introduce a native latent world-action model.
Following its large-scale human video pretraining roadmap, BeingBeyond has successively released the Being-H general-purpose dexterous manipulation model and the Being-M general-purpose mobile manipulation model, scaling pretraining to 500,000 hours of human video data. The company has also built a full-stack infrastructure spanning data pipelines, model pretraining, post-training, evaluation, and on-device deployment. To date, BeingBeyond has established ecosystem partnerships with dozens of industry players, including Linkerbot, AI² Robotics, and D-Robotics, continuously advancing the deployment of general-purpose embodied models across different robot embodiments and tasks.
Daimon Robotics is dedicated to providing the key capabilities required for stable robotic operation under real physical contact and in complex environments. As the world's first "physical interaction brain" designed to power dexterous robotic manipulation, the Daimon-TWM tactile-grounded world model integrates physical cognition, predictive decision-making, and real-time control within a unified framework. By using real tactile feedback for physical reasoning, the model can anticipate contact evolution and potential failure risks in advance, and perform millisecond-level corrections before errors occur. It has delivered up to a 10× improvement in task success rates, while demonstrating strong disturbance robustness and cross-platform generalization.

BeingBeyond's embodied foundation models provide robots with high-level task understanding and planning capabilities, while Daimon-TWM enhances rapid response and corrective control during task execution. The complementary capabilities of the two companies enable robots not only to understand tasks and their environments, but also to sense changes and adjust their actions after making physical contact with objects, further improving adaptability, stability, and precision in complex manipulation tasks.
For real-world deployment, the two companies will initially focus on 3C precision manufacturing and flexible production, conducting POC validation around high-precision dexterous manipulation tasks such as deformable-object handling, precision assembly, and adaptive insertion and unplugging. Together, they will advance the deployment of embodied models and tactile capabilities in real production environments, creating tangible industrial value.
Through cross-embodiment adaptation and full-stack on-device deployment, the partnership will bridge the critical path from capability validation to scalable replication, enabling physical intelligence to flow across robot embodiments, workstations, and production lines, and establishing a general-purpose intelligent foundation for the next generation of precision and flexible manufacturing.
About BeingBeyond
BeingBeyond focuses on developing and applying general-purpose embodied foundation models. It was among the first companies in China to propose a general-purpose model framework trained on large-scale human video data, and the first AI startup in China to introduce a natively latent-space world model. With a mission to bring humanoid robots out of the laboratory and into everyday life, BeingBeyond is committed to solving the core technical challenges of embodied AI and leading the transformation of humanoid robotics.