DeepRoute.ai showcased its frontier Vision-Language-Action (VLA) model at the 2025 Seoul Mobility Show last week, marking a significant milestone in its global expansion strategy. Designed for versatility, the VLA model supports camera-only and lidar-camera fusion configurations. It is compatible with various chip platforms, and the company has confirmed it plans to integrate the VLA model into more than five vehicle models this year alone.
The first vehicle integrated with DeepRoute.ai’s VLA model – the first of its kind in China – is set for release to the consumer market in mid-2025. This advanced system excels in long-context understanding, analyzing driving scenarios over extended timeframes of up to a dozen seconds, according to the company. It provides step-by-step explanation of its decision-making process, offering insights into its surroundings, predictions and planned actions.
“South Korean auto makers have been at the forefront of smart driving innovation, rapidly integrating the latest technologies into mass-production models,” said Maxwell Zhou, CEO of DeepRoute.ai. “With our extensive experience in scaling smart driving vehicles for the Chinese market, we look forward to collaborating closely with South Korean manufacturers to advance AI-powered mobility.”
Founded by Zhou in 2019, the company is headquartered in Shenzhen, with businesses across the globe. DeepRoute.ai has completed six rounds of funding, accumulating over US$500m. The company aims to create artificial general intelligence using series production vehicles and large-scale robotaxi deployment to collect a vast amount of data from the physical world. This data will enable DeepRoute.ai to build an AI brain with “human-level wisdom” to help boost productivity in various industries.