Moore Threads: Yumetech successfully completed the full-process training of the self-developed large brain model RoboBrain 2.5.

date
13/01/2026
Recently, Moore Threads, in collaboration with the Beijing Intelligent Source Artificial Intelligence Research Institute, successfully completed the full-process training of the self-developed RoboBrain 2.5 large-scale brain model based on the FlagOS-Robo framework and relying on the MTT S5000 K-Calculus cluster. This is the first time in the industry to verify the availability and efficiency of domestic computing clusters in the training of large-scale intelligent models, marking a crucial step for domestic AI infrastructure in addressing complex multimodal tasks.