PATEO Integrates NVIDIA and ByteDance AI Models into Automotive Cockpit Systems
PATEO is deploying NVIDIA Nemotron and ByteDance Doubao language models within its NVIDIA DRIVE AGX Thor-based AI Box to provide local, real-time in-vehicle processing.
PATEO has announced the integration of NVIDIA’s Nemotron speech recognition and ByteDance’s 35-billion-parameter Doubao language model into its automotive AI platform. The system, built on the NVIDIA DRIVE AGX Thor architecture, performs inference locally to facilitate multilingual voice interaction, cabin context awareness, and vehicle control without requiring a constant cloud connection.
Demonstrated at Auto China 2026, the technology is being developed alongside ByteDance and an unnamed electric vehicle manufacturer. PATEO confirmed a project nomination for a new vehicle model, with mass production scheduled to commence in the second quarter of 2027.
The system utilizes a modular software architecture that allows for the addition of high-performance AI capabilities to existing vehicle platforms without replacing the primary control hardware. By processing sensor, visual, and audio data on-device, the company aims to improve reliability and reduce latency, particularly in regions with unstable network infrastructure.
PATEO is also expanding its business model beyond initial hardware sales by developing a token-based ecosystem for ongoing AI services. Partnering with firms like Mininglamp Technology, the company is building a foundation for broader physical AI applications, including potential entry into the robotics sector.