PATEO integrates NVIDIA and ByteDance AI models for in-vehicle computing
PATEO has announced an AI box powered by NVIDIA DRIVE AGX Thor to run large language models and edge computing within vehicles.
Shanghai-based PATEO has introduced an AI-driven computing unit that integrates NVIDIA Nemotron technology and ByteDance’s 35-billion-parameter Doubao language model. Built on the NVIDIA DRIVE AGX Thor platform, the hardware is designed to perform localized AI inference, enabling vehicles to process complex voice commands and sensor data without requiring constant cloud connectivity.
The system coordinates cabin functions and interprets user intent by combining voice, visual, and sensor inputs. By running these processes locally, the platform aims to provide low-latency performance for features such as real-time language translation and proactive driver assistance, including fatigue detection and climate control management.
At the Auto China 2026 exhibition, PATEO confirmed that the solution is being developed alongside ByteDance and an unnamed electric vehicle manufacturer, with mass production scheduled for the second quarter of 2027. The company intends for the AI box to serve as a scalable module, allowing existing vehicle architectures to gain advanced AI capabilities without the need to replace their primary control processors.