Core Insights - The article discusses the development of the EmbodiedBrain model by ZTE NebulaBrain Team, which aims to address the limitations of current large language models (LLMs) in embodied tasks, focusing on robust spatial perception, efficient task planning, and adaptive execution in real-world environments [2][4]. Group 1: Model Architecture - EmbodiedBrain utilizes a modular encoder-decoder architecture based on Qwen2.5-VL, achieving an integrated loop of perception, reasoning, and action [5]. - The model processes various multimodal inputs, including images, video sequences, and complex language instructions, generating structured outputs for direct control and interaction with embodied environments [8][10]. - Key components include a visual transformer for image processing, a lightweight MLP for visual-language integration, and a decoder that enhances temporal understanding of dynamic scenes [9][10]. Group 2: Data and Training - The model features a structured data architecture designed for embodied intelligence, ensuring alignment between high-level task goals and low-level execution steps [12]. - Training data encompasses four core categories: general multimodal instruction data, spatial reasoning data, task planning data, and video understanding data, with a focus on quality through multi-stage filtering [14][15]. - The training process includes a two-stage rejection sampling method to enhance model perception and reasoning capabilities, followed by a multi-task reinforcement learning approach called Step-GRPO to improve long-sequence task handling [20][21]. Group 3: Evaluation System - EmbodiedBrain establishes a comprehensive evaluation system covering general multimodal capabilities, spatial perception, and end-to-end simulation planning, addressing the limitations of traditional offline assessments [26][27]. - The model demonstrates superior performance in various benchmarks, including MM-IFEval and MMStar, indicating its enhanced multimodal capabilities compared to competitors [28][29]. - In spatial reasoning and task planning evaluations, EmbodiedBrain achieves significant improvements, showcasing its ability to perform complex tasks effectively [30][31]. Group 4: Case Studies and Future Outlook - The model successfully executes tasks involving spatial reasoning and end-to-end execution, demonstrating its capability to generate coherent action sequences based on complex instructions [37][41]. - ZTE plans to open-source the EmbodiedBrain model and its training data, aiming to foster collaboration in the field of embodied intelligence and address existing challenges in data accessibility and evaluation standards [42][43]. - Future developments will focus on multi-agent collaboration and enhancing adaptability across various real-world robotic platforms, pushing the boundaries of embodied intelligence applications [43].

12月3日消息，跌近4%，报32.46港元，成交额12.05亿港元。 摩根士丹利最新研报中表示，尽管豆包演示版本展现了丰富的功能生态系统，但大摩认为其落地执行面临重大挑战。报告指出，这种深度的系统级整合需要 修改操作系统，这直接触动了手机厂商（OEM）的利益核心。大摩认为，主要手机厂商和超级应用更倾向于自研AI助手。 编辑/melody 消息面上，据科创板日报报道，记者查询获悉，在中兴商城官网，售价3499元的nubia M153 豆包手机助手技术预览版已经售罄。中兴终端方面表示，nubia M153手机是合作的工程样机，由豆包主导AI手机助手的产品定义与体验，中兴通讯旗下nubia主导硬件工程与技术研发。仅面向需要体验豆包手机的行业人 士，展示豆包AI技术与能力，非大众消费级产品。 ...

Core Viewpoint - ByteDance and ZTE have launched the Doubao Phone Assistant, which has gained significant attention and led to a substantial increase in ZTE's market value, highlighting the potential of AI in mobile technology [2][3]. Group 1: Product Launch and Market Reaction - The Doubao Phone Assistant was released on December 1, with the ZTE nubia M153 model priced at 3499 CNY selling out quickly [2][3]. - ZTE's stock surged by 20 billion CNY in market value on the day of the launch, indicating strong investor interest in AI-driven mobile solutions [2][3]. Group 2: Strategic Partnerships and Future Plans - Currently, ZTE is the only mobile manufacturer collaborating with Doubao, but ByteDance plans to partner with multiple manufacturers to integrate the assistant into various devices [3][4]. - The strategy aims to establish a foothold in the AI-driven human-computer interaction space, with mobile devices being a primary focus for AI applications [3][4]. Group 3: Industry Trends and Challenges - The trend in the mobile industry is shifting towards integrating AI assistants as central components of the user experience, moving from standalone features to ecosystem hubs [4][5]. - Challenges include data privacy concerns and the need for AI systems to be accepted by mobile operating systems to function effectively, as demonstrated by user issues with WeChat integration [5].

Core Viewpoint - Users of Nubia M153 reported issues with WeChat crashing when using Doubao Phone Assistant, raising concerns about the software's functionality and security [1][2][3] Group 1: Product Overview - Nubia M153 is an AI phone developed by ZTE in collaboration with ByteDance, featuring the Doubao Phone Assistant at a price of 3499 yuan [1][2] - Doubao Phone Assistant is an AI software that enhances user interaction and experience, built on the Doubao APP and developed in partnership with mobile manufacturers [1][2] Group 2: Technical Issues - Users confirmed that the Doubao Phone Assistant caused WeChat to crash or become inaccessible during operation [1][2] - The engineering version of Doubao Assistant can currently use WeChat, but the functionality to operate WeChat has been disabled [3] Group 3: Permissions and Security - Doubao Assistant has the INJECT_EVENTS permission, which allows it to simulate user input events, categorized as a high-risk permission in Android [3] - Doubao stated that the permissions are disclosed in their privacy policy and that the assistant's operations require user authorization and supervision [2][3] - Industry insiders noted that INJECT_EVENTS is a common permission for mobile assistants, with other brands like Xiaomi and Honor also utilizing it [4]

Market Overview - The market experienced fluctuations throughout the day, with the ChiNext Index dropping over 1% after initially rising more than 1% in the morning [1] - By the end of the trading session, the Shanghai Composite Index fell by 0.51%, the Shenzhen Component Index decreased by 0.78%, and the ChiNext Index declined by 1.12% [1] - More than 3,800 stocks in the market closed lower [1] Sector Performance - The superhard materials sector showed strong performance, with Huanghe Whirlwind hitting the daily limit [1] - The coal sector also performed well, with Antai Group achieving two consecutive limit-ups, and Dayou Energy and New Dazhou A both hitting the daily limit [1] - The wind power sector saw gains, with Dajin Heavy Industry and Delijia both reaching the daily limit [1] - The commercial aerospace concept surged in the afternoon, with Shunhao Co. achieving four consecutive limit-ups and Aerospace Machinery Electric hitting the limit twice in four days [1] Declining Sectors - The AI application sector collectively weakened, with Fushi Holdings and Yinsai Group experiencing significant declines [2] Trading Volume - The total trading volume for the Shanghai and Shenzhen markets reached 1.67 trillion yuan, an increase of 76.5 billion yuan compared to the previous trading day [3] Individual Stock Highlights - Xinyi Sheng had a trading volume exceeding 22.1 billion yuan, leading the market [4] - Other notable stocks by trading volume included Zhongji Xuchuang, ZTE Corporation, and Aerospace Development [4]

Core Points - The total trading volume of the Shanghai and Shenzhen stock markets reached 1.67 trillion yuan on December 3, an increase of approximately 765.32 billion yuan compared to the previous trading day [1] - The Shanghai stock market accounted for 647.17 billion yuan, while the Shenzhen stock market contributed 1.02 trillion yuan to the total trading volume [1] Company Highlights - New Yisheng led the trading volume with 22.10 billion yuan [1] - Other notable companies by trading volume included Zhongji Xuchuang at 19.82 billion yuan, ZTE Corporation at 13.40 billion yuan, Aerospace Development at 12.47 billion yuan, and Tianfu Communication at 11.78 billion yuan [1]

Core Insights - The discovery of Smartisan remnants in the Doubao mobile system indicates a connection to the legacy of Smartisan Technology, suggesting a blend of past innovations with new developments in mobile technology [1][3]. Group 1: Company Background - Doubao mobile system is developed in collaboration between ZTE and ByteDance, with ByteDance leading the AI mobile assistant product definition and experience, while ZTE's Nubia handles hardware development and manufacturing [1]. - Smartisan Technology, founded in June 2017 with a registered capital of 200 million yuan, has shifted its focus after the departure of its founder, Luo Yonghao, who has since ventured into AI startups [1]. Group 2: Product Features - The Doubao mobile system includes classic Smartisan ringtones, such as "Mi Store," indicating a nostalgic element for users familiar with Smartisan products [1][3]. - The first phone featuring the Doubao mobile assistant technology is the Nubia M153, showcasing a blend of AI capabilities and traditional mobile features [1]. Group 3: Leadership Changes - Luo Yonghao stepped down as the executive director and general manager of Smartisan Technology in September 2022, with Guan Zhiliang taking over the role [1]. - Luo has launched a new AI startup project named Jarvis, which went live overseas earlier this year [1].

现在的AI 如果不被手机操作系统接纳，就只能是一个聊天机器人玩具；只有进入系统底层，才能成为 真正的"助手"。 在AI的时代，传统的APP孤岛会被彻底击穿。当然，这条路注定充满荆棘。这背后更 深层次的问题是权限。 豆包官方表示，在手机助手的落地路径上，没有自研手机计划，目前正与多家手机厂商洽谈，以"生态 合作"形式将豆包手机助手整合进不同品牌机型中，相关进展将在后续公布。 这一战略背后，是字节跳动对AI时代人机交互入口的全面争夺。字节正在积极探索硬件入口，以手机 为代表的智能终端，是AI最直接的落地场景。 当下的主流手机厂商纷纷自研端侧AI，比如苹果的Siri、三星的Bixby、华为的小艺、小米的小爱。今年 10月份vivo、OPPO、荣耀扎堆发新品，手机AI智能体成为了主打卖点。 主流手机厂商这种"生态系统闭环"策略，让大模型厂商急需找到属于自己的端侧入口，集成和释放AI能 力。豆包此次选择以"AI手机助手"形态切入，正是看准了这一点。 2025年9月，手机厂商AI助手用户规模达到了5.35亿，这是检验上游端侧模型技术的"试金石"。当下， 手机AI助手的行业趋势是从"功能模块"升级为"生态中枢"，通过系统级 ...

Core Insights - The "Doubao Assistant" phone, priced at 3499 yuan, is sold out on ZTE's official website [1] - On the second-hand market (Xianyu), the resale prices for the phone exceed the original price, with new, unopened units listed between 7999 to 9999 yuan, indicating a premium of over 6500 yuan [4] Group 1: Product Performance - The "Doubao Assistant" phone has experienced high demand, leading to its sell-out status on the official platform [1] - Resale prices on Xianyu suggest strong market interest and potential for profit among resellers [4] Group 2: User Experience Issues - Users of the Nubia M153 engineering model reported issues with logging into WeChat, indicating potential compatibility problems with the app [6] - The phone's assistant feature fails to perform tasks involving WeChat, as it is currently unsupported [7] - A response from WeChat representatives suggests that the issues may be related to security measures rather than a technical fault [10]

Group 1 - The core point of the news is the launch of the Doubao Mobile Assistant by ByteDance in collaboration with ZTE, which aims to enhance user interaction and experience through AI capabilities [1][2]. - Users of the Doubao Mobile Assistant reported issues with logging into WeChat, indicating potential security restrictions from WeChat [1]. - The Doubao Mobile Assistant is currently available for developers and tech enthusiasts on the Nubia M153 prototype, which has already sold out on ZTE's official store [2]. Group 2 - The second-hand market for the Nubia M153, which features the Doubao Assistant, shows prices ranging from 4,200 to 4,999 yuan, exceeding the official retail price by 700 to 1,500 yuan [2]. - WeChat representatives stated that there were no special actions taken that would affect users, suggesting that the issues may be related to standard security measures [1].