Core Insights - SenseTime participated in the 2025 World Intelligent Industry Expo, showcasing its core products including the multi-modal large model, AIOS, and New Member [1] - The partnership between SenseTime and Changan Automobile has been stable since 2022, with nearly 2 million vehicles delivered across 20 models utilizing SenseTime's DMS/OMS products [1] - The multi-modal large model integrates language understanding, full-scene perception, multi-modal fusion, and deep reasoning to enhance user interaction in vehicles [1] Group 1 - SenseTime's multi-modal large model aims to transform vehicle interaction into a more natural and fluid communication experience [1] - The model has been deployed on mainstream vehicle chips, creating a complete gradient from lightweight to flagship products [1] - The New Member feature enables wake-word-free multi-party conversations, allowing it to understand and participate in discussions without interrupting [2] Group 2 - The AI kernel, "Jianji," is designed specifically for vehicle AIOS, integrating various AI capabilities to enhance autonomous perception and decision-making [2] - New Member possesses multi-modal recognition and deep thinking capabilities, transitioning from passive response to proactive service [2]

Group 1 - The automotive industry is undergoing a significant transformation driven by AI large models and multimodal interaction technology, enabling smart cockpits to not only display driving information and navigation but also engage in communication and assist with tasks like ordering food and booking tickets [1] - The 2025 World Artificial Intelligence Conference (WAIC 2025) showcased the latest technological advancements in smart cockpits, highlighting the shift towards proactive intelligence in automotive applications [1][2] - The implementation of edge-based multimodal large model solutions can achieve a 90% service closure in perception, decision-making, and execution within smart cockpits, enhancing user experience through natural dialogue [2] Group 2 - The core challenge of multimodal interaction in practical scenarios lies in the fusion of multimodal signals and prioritization, with dynamic weight algorithms adjusting sensitivity based on situational complexity [2] - The IDC report indicates that vehicle-side interaction is evolving from "command execution" to "proactive service," becoming a key competitive technology for automakers [2] - The design of smart cockpit systems aims to create products with human-like intelligence, capable of understanding, remembering, and interacting proactively, thus transforming the cockpit into a third living space for users [3] Group 3 - The integration of smart cockpit and smart driving technologies is becoming increasingly urgent, with a focus on ensuring compliance in technology, data, and safety to mitigate regulatory risks [5] - The adoption rate of smart cockpits in passenger vehicles priced over 200,000 yuan is projected to reach 100% by 2024, with significant increases in lower-priced segments as well [4] - The automotive industry is encouraged to view regulatory compliance not merely as a burden but as an opportunity for product innovation and market differentiation [5]