XPeng Group announced a major milestone: its first mass-produced pure vision Robotaxi rolled off the production line in Guangzhou, marking the first time a Chinese automaker has completed front-loaded mass production of a Robotaxi through full-stack in-house R&D. Based on XPeng’s flagship GX model, the vehicle is equipped with 4 self-developed Turing AI chips delivering a total computing power of 3,000 TOPS—the highest in the global automotive industry—paired with XPeng’s second-generation VLA large model for Level 4 autonomous driving capability.

Unlike most Robotaxis, XPeng’s model relies neither on LiDAR nor high-precision maps, using only a pure vision solution combined with the VLA large model for decision-making. The new model eliminates the language translation step in traditional architectures, compressing system response latency to less than 80ms and significantly improving generalization ability, laying a solid foundation for XPeng’s future global Robotaxi expansion.