XPeng: End-to-end large model training volume exceeds 20 million clips, the only one in the country, and will reach a scale of 100 million clips by the end of 2025

Recently, at the XPeng Turing AI Autonomous Driving Open Day, XPeng officially stated that starting from the XPeng P7+, there will no longer be a distinction between Max and Pro, and AI autonomous driving will become a standard feature regardless of vehicle model or hardware configuration. XPeng also mentioned that by 2025, it aims to provide users with a better autonomous driving experience through a one-piece end-to-end solution, achieving less than one takeover per 100 kilometers. Yuan Tingting, Senior Director of Autonomous Driving Products at XPeng, stated that the XPeng AI Tianji 5.5.0 system is already capable of driving from parking space to parking space without interruptions, and it supports intelligent exit with left and right turns. XPeng has also become the first company in the industry to achieve "parking space to parking space" driving with a single software solution, integrating underground parking, gate systems, internal roads of parks, and city roads. "Autonomous driving has become the primary reason for purchasing the XPeng P7+, with the penetration rate of high-level intelligent driving in China reaching about 10%," she said. "XPeng's end-to-end large model training volume is the only one in the country exceeding 20 million clips, and it will reach a scale of 100 million clips by the end of 2025."