BeingBeyond and PaXini Tech Enter Comprehensive Partnership

BeingBeyond, a leading developer of general-purpose embodied foundation models, recently signed a cooperation agreement with PaXini, a global leader in embodied perception-driven general intelligence.

The two companies will work together across embodied foundation-model research and development, proof-of-concept validation in deployment scenarios, and the joint development of embodied-AI data standards. The partnership aims to expand the embodied AI market.

BeingBeyond Co-founder Li Zhang (left) and PaXini Co-founder Hengdi Zhang (right) sign the strategic cooperation agreement.

The partnership brings together BeingBeyond's capabilities in general-purpose embodied foundation models and PaXini's strengths in high-precision multidimensional tactile perception and large-scale data. The companies will collaborate across embodied-model training, proof-of-concept validation in real-world settings, and model and algorithm optimization.

BeingBeyond was among the first companies in China to propose pretraining general-purpose embodied models on large-scale egocentric human video. The company has released Being-H, a general-purpose foundation model for dexterous manipulation that supports real-time on-device deployment, and Being-M, a general-purpose foundation model for mobile manipulation. On July 28, 2026, BeingBeyond released Being-H0.8, its latest latent tactile world-action model. Pretrained on more than 500,000 hours of egocentric human video, the model brings tactile signals into large-scale model pretraining, enabling robots to develop a more complete understanding of physical interaction and better predict how those interactions unfold. PaXini, one of BeingBeyond’s data partners, provided high-quality multimodal data for this work.

PaXini has developed OmniSharing DB, an omni-modal embodied intelligence dataset that spans vision, touch, force sensing, action trajectories, and other core modalities. It draws on the company's multidimensional tactile sensors, which rank first worldwide by shipment volume, and its Human-Centered system for collecting real-world omni-modal perception data. The dataset captures human hand motions with up to 82 degrees of freedom and integrates 30 six-axis tactile modules. Its industrial-grade temporal and spatial alignment operates at millisecond-level precision, allowing the data to be used directly for embodied-model training without complex post-processing.

PaXini has established a cluster of large-scale data factories across China. Its end-to-end data pipeline covers sensing hardware, data collection, cleaning and annotation, quality validation, and scaled delivery. With capacity approaching ten billion omni-modal data points, PaXini can respond quickly to both large-volume foundation-data requirements and customized data needs for long-tail scenarios, while maintaining industrial-grade standards.

The partnership extends beyond technology and business development to the joint work on industry standards. On August 14, 2026, the Workshop on Visual Perception Data Standards for Humanoid Robots and Embodied AI was held in Beijing. BeingBeyond hosted the event. As a co-lead organization of the WG7 Data Working Group under the Ministry of Industry and Information Technology's Standardization Technical Committee for Humanoid Robots and Embodied Intelligence, PaXini participated extensively in the workshop. The companies' close coordination on data standards laid a strong ecosystem foundation for this comprehensive partnership.

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.