Group 1 - Quark launched a new AI creation platform called "ZaoDian" on September 24, integrating AI image and video generation capabilities, and is the first platform in China to support synchronized audio-visual video generation [1] - "ZaoDian" utilizes the leading image generation model Midjourney V7, combined with Quark's proprietary technology, providing a one-stop creative experience for users [1] - The advancement of AI technology and its expanding application scenarios indicate a promising future for AI creation beyond just images and videos [1] Group 2 - AI startup "Mianbi Intelligent" announced the debut of its automotive business line on September 24, with a multimodal large model (0.9B) developed in collaboration with Geely Central Research Institute [2] - The model enhances the stability and reliability of the cockpit system, featuring multimodal perception integration, real-time response, and proactive decision-making capabilities, enabling functions like intelligent fog lights and adaptive windows [2] - This collaboration exemplifies the deep application of AI technology in the automotive sector, highlighting its potential in traditional industries such as manufacturing and transportation [2] Group 3 - Alibaba announced a partnership with NVIDIA on September 24 during the "2025 Cloud Habitat Conference," focusing on Physical AI collaboration across various aspects including data synthesis, model training, environmental simulation reinforcement learning, and model validation testing [3] - The collaboration between Alibaba and NVIDIA represents a significant strategic move in the field of artificial intelligence, aiming to propel the development of intelligent agents and robotics technology [3]

Core Viewpoint - The establishment of the automotive business line by the AI startup, Mianbi Intelligent, has led to the successful deployment of a multimodal large model (0.9B) in the Geely Galaxy M9 smart cockpit platform, enhancing system stability and reliability [1] Group 1: Company Developments - Mianbi Intelligent has collaborated with Geely Central Research Institute to develop the edge-side VLA multimodal large model, which integrates multimodal perception and real-time response capabilities [1] - The model enables features such as intelligent fog lights and adaptive windows, while local data processing reduces the risk of user information leakage [1] - In July, Mianbi Intelligent announced a new organizational upgrade to establish its automotive business line, aiming for a breakthrough in applying its MiniCPM edge-side model to more vehicles [1]