XPENG officially named its robotaxi business "XPENG YOYO" on October 8, 2026 and opened autonomous-driving online registration to the public in China. Registration runs through invitation codes, with a dedicated YOYO page now live on the official website. The move shifts the program from technical validation to marketization and industrialization, which matters for companies watching how fast robotaxi supply becomes commercially available.
How YOYO moves from testing to commercial brand
YOYO will operate with a defined brand identity and commercial positioning, providing operation services to users and partners. XPENG calls the vehicle China's first fully in-house developed robotaxi to achieve pre-loaded mass production. The program rolled off the production line in Guangzhou in May 2026, then completed the full chain from online ride-hailing to automated pickup and drop-off in July 2026. In August 2026 it obtained Guangzhou's remote testing qualification for intelligent connected vehicles, allowing road tests without a safety operator in the driver's seat on Class I, II and III test roads.
The vehicle platform combines four in-house Turing AI chips delivering 3,000 TOPS, described by XPENG as the world's highest in-vehicle compute. It runs on the VLA2.0 large model and operates without high-definition maps or LiDAR, a design intended to allow rapid deployment across cities and environments. The official page repeats that the system does not depend on high-definition maps and adapts quickly in a new city. XPENG says this path bridges L2 intelligent driver assistance and L4 autonomous driving, extending AI capabilities from its automotive business in a way it characterizes as entirely different from traditional L4 companies.
XPENG selected the English name YOYO for recognition in global markets, citing short length, easy pronunciation and consistency across language environments. The dedicated page presents YOYO as an L4 autonomous-driving solution built for global use, integrating vehicle platform, Turing chips, second-generation VLA model and intelligent manufacturing into one platform meant to scale and keep evolving. Partners can connect quickly without building an autonomous-driving system from scratch and shape mobility services around local markets and operating scenarios. The company also frames the program as a key practice of its Physical AI strategy, bringing AI from the digital world into the physical world through the automobile.
What YOYO means for corporate mobility buyers
For operators and corporate users, the partner model changes procurement logic: XPENG positions itself as technology provider and ecosystem enabler, with revenue from hardware sales, services and commission sharing, while offline operations go to global partners. A small mobility firm could therefore launch a branded service without developing autonomy in-house, focusing resources on fleet, local permits and customer service. A large transport group or platform could add YOYO as another vehicle and software stack alongside existing suppliers, testing automated pickup and drop-off on defined routes before wider rollout.
The offer still carries checks that matter before any commitment. Before public opening, XPENG ran paid internal testing at 1 renminbi per ride to validate payment and service workflow, so external validation in real scenarios is only starting. Cabin features such as temperature preset, fragrance control, zero-gravity seats with ventilation, heating and massage, voice assistant and entertainment screen are confirmed, but service coverage, pricing and safety performance outside Guangzhou are not. Buyers should ask about city approvals, remote supervision, data handling, maintenance of Turing-chip vehicles and division of liability between XPENG and the local operator.
The marker to follow is operational validation in Guangzhou during 2026 under L4 demonstration-operations qualification, followed by a new-generation robotaxi model planned for 2027 and global deployment at scale after that. If employee testing expands into sustained external rides and the 2027 model appears on schedule, the shift from single-point validation to scaled deployment will gain weight for fleet planning.
