SenseAuto was featured at NVIDIA’s Exhibition Booth, Showcasing Significant Achievements in the Deployment of Edge Multimodal Large Models

Recently, the 3rd China International Supply Chain Expo (CISCE) kicked off in Beijing. NVIDIA, a global leader in artificial intelligence, made its debut at the expo, where SenseAuto was showcased at NVIDIA’s booth.

Leveraging NVIDIA’s high-performance chip platform, SenseAuto successfully demonstrated its technological innovations and outstanding achievements in deploying edge multimodal large models.

As a key ecosystem partner of NVIDIA in automotive AI, SenseAuto has conducted in-depth development and innovation on NVIDIA’s high-performance computing platform, achieving breakthroughs in critical technologies such as operator development, model quantization, and precision support. These advancements have led to a GPU utilization increase from 35% to 85%, a 4x boost in low-precision computing speed, and extended visual-text support, delivering a natural and seamless interactive experience to users.

At this year's CISCE, SenseAuto showcased its edge multimodal large model, built upon SenseTime's SenseNova model, the top-ranked multimodal reasoning model in China, and specifically customized for smart cockpit scenarios. Equipped with full-scene multimodal perception, comprehension, and reasoning capabilities, it accurately perceives and deeply understands user needs along with in- and out-of-vehicle environments. This enables the delivery of highly personalized and proactive services, transforming smart vehicles into users' new assistants, companions, and even becoming extended family members.

In November 2024, SenseAuto launched "A New Member For U" (New Member), powered by groundbreaking AI technologies including its native streaming multimodal large model. This innovation enables vehicles to "perceive moods, remain ever-present, and connect with you intuitively," giving every car "a soul with personality". In January 2025, New Member took its first developmental leap: SenseAuto partnered with leading Chinese automakers to deploy the industry’s first native streaming multimodal cockpit system into mass production. Current production features include preceding vehicle detection, traffic sign recognition, scenic route identification, in-cabin object recognition, occupant monitoring, and a proactive greeting system.

At NVIDIA GTC 2025 in March this year, Dr. Wang Xiaogang, Co-founder, Executive Director and Chief Scientist of SenseTime, was invited to deliver a keynote speech titled "Unleashing the Creativity of AGI: Pioneering a New Future for Smart Vehicles." In his address, he shared SenseAuto's innovative autonomous driving solutions developed on the NVIDIA Drive AGX platform.

As smart vehicles ride the wave toward the Artificial General Intelligence (AGI) era, on-device multimodal large models are becoming the crucial nexus. SenseAuto will continue collaborating with industry partners to pioneer more AI innovations and products, building a new smart cockpit ecosystem to accelerate the arrival of AGI-powered smart vehicles.

In December 2024, SenseTime completed its transition to a "1+X" organizational structure. As a key "X" ecosystem company of SenseTime, SenseAuto focuses on accelerating smart vehicles' transition into the AGI era. We are committed to delivering future mobility solutions that are adaptively flexible, deeply personalized, trustworthy and human-centric.