Gasgoo Munich-SYNAPATH AI and BeingBeyond have officially signed a strategic partnership agreement. The two companies will deepen collaboration on high-quality data development, model training validation, and industry data standards, while exploring the creation of a data-sharing ecosystem platform for the sector.
As BeingBeyond’s core data partner, SYNAPATH AI has already contributed to its data ecosystem, providing high-quality data support for training embodied foundation models. During the development and training of the Being-H0.8 implicit tactile world action model, Shutu’s data capabilities were integrated directly into the R&D pipeline and validated in real-world model applications.
Building on their work in model development, the companies are now extending their partnership into data standardization and quality system construction.
As an infrastructure provider in the Physical AI sector, SYNAPATH AI focuses on delivering reusable human behavior data products and technical capabilities tailored to various models, robot bodies, and application scenarios.
Image source: SYNAPATH AI
SYNAPATH AI has established a complete production workflow spanning task design, behavior capture, data governance, intelligent structuring, and customized delivery. With industrial-scale delivery capabilities for model training and real-world robot verification, the company already serves top-tier embodied AI firms, internet giants, and traditional robotics manufacturers.
BeingBeyond, meanwhile, specializes in the research and application of general embodied foundation models. Since its inception, the company has hit several key milestones centered on "large-scale human video pre-training," launching the world’s first embodied foundation models trained on thousands of hours, over 10,000 hours, 200,000 hours, and 500,000 hours of human video data.
BeingBeyond has now built a full-stack infrastructure covering data pipelines, pre-training, post-training, evaluation, and edge deployment. This setup enables the continuous accumulation of large-scale data and the ongoing evolution of model capabilities.
