Core Insights - Baidu has established two new core departments: the Basic Model Research Department and the Application Model Research Department, both reporting directly to CEO Li Yanhong, indicating a strategic focus on AI development [3][5]. Group 1: Department Functions - The Basic Model Research Department, led by Wu Tian, will focus on developing high-intelligence, scalable general AI models, supporting all business lines with foundational AI technology [3][4]. - The Application Model Research Department, headed by Jia Lei, will specialize in tuning and exploring models tailored to specific business application scenarios, integrating general model technology with actual business needs [4]. Group 2: Leadership and Structure - Both new departments report directly to Li Yanhong, breaking from the previous structure where the model research department reported to the CTO, reflecting Baidu's heightened emphasis on large model business [5]. - Wu Tian and Jia Lei are both seasoned leaders with extensive experience in AI, having made significant contributions to Baidu's core products [3][4]. Group 3: Financial Performance - Baidu's Q3 2025 financial report showed total revenue of 31.2 billion yuan, a year-on-year decline of 7% and a quarter-on-quarter decrease of 5% [5]. - The company reported a net profit of 3.77 billion yuan, down 36% year-on-year and 21% quarter-on-quarter, with a net loss of 11.22 billion yuan [5]. - AI business revenue was disclosed for the first time, totaling 9.6 billion yuan, with a year-on-year growth of over 50%, and AI cloud revenue increased by 33% to 4.2 billion yuan [6]. Group 4: Strategic Implications - The restructuring and focus on AI core business suggest that Baidu aims to allocate more resources to enhance its AI capabilities and address potential shortcomings in current product performance [7].

Core Insights - The event "Wenxin Mentor Training Camp - 'Finance + AI Data' Special" will be held on December 6-7, 2025, in collaboration with Baidu Wenxin and Beijing News, aiming to promote the deep application of AI technology in the financial sector and cultivate interdisciplinary talents [1] Group 1: Event Overview - The training camp responds to national initiatives, specifically the "14th Five-Year Plan," which emphasizes the implementation of "AI+" actions to integrate AI with various industries [1] - The camp focuses on accelerating the application of large model technology in finance, featuring expert-led courses and real case studies [1][2] - Participants will gain practical training that combines theory with real-world applications, enhancing their financial business insights and AI technical skills [1] Group 2: Training Content - The curriculum includes techniques for maximizing AI potential, such as "Financial Prompt Engineering," and methods for evaluating AI responses across financial scenarios [2] - The training camp is open for registration to individuals with at least three years of frontline financial experience or those holding a PhD in finance or related fields [3] Group 3: Networking Opportunities - Participants will have the opportunity to learn cutting-edge methodologies for cross-disciplinary integration and join a high-quality fintech network [4]

Group 1 - The core viewpoint of the news highlights the strong performance of the Sci-Tech Innovation Artificial Intelligence ETF (589010), which rose by 0.68% and showed resilience despite some individual stock pullbacks [1] - Key holdings in the ETF include Lexin Technology and Chip Original Co., both of which increased by over 5%, while leading stock Hanwei Technology rose nearly 3% [1] - Baidu announced the establishment of a new technical research organization, including a foundational model research department and an application model research department, both reporting to CEO Li Yanhong [1] Group 2 - The Sci-Tech Innovation Artificial Intelligence ETF closely tracks the Shanghai Stock Exchange Sci-Tech Innovation Board AI Index, covering high-quality enterprises across the entire industry chain [2] - The ETF benefits from high R&D investment and policy support, with a 20% price fluctuation limit and small-cap elasticity aiding in capturing the "singularity moment" of the AI industry [2]

作为百度AI商业化落地的重要业务板块，慧播星数字人直播是近年来的明星产品，其背后正是吴甜团 队提供的核心技术支持。 【百度新设两个大模型研发部 直接向CEO李彦宏汇报】11月25日，百度发布公告，将新设立两个大模 型研发部门，其中包括新设基础模型研发部，负责研发高智能可扩展的通用人工智能大模型，由吴甜负 责；新设立应用模型研发部，负责业务应用场景需要的专精模型调优和探索，由贾磊负责。王海峰继续 担任百度CTO、TSC主席、百度研究院院长。以上几个技术部门均向百度CEO李彦宏汇报。 百度百科显示，吴甜于2006年加入百度，在百度先后担任过搜索RANK及其他部门负责人、百度AI技术 平台体系执行总监、深度学习技术及应用国家工程实验室副主任等职位，现任百度集团副总裁。 在百度的近20年时间，吴甜先后带领团队在机器学习、深度学习、数据挖掘、自然语言处理、计算机视 觉、增强现实等人工智能重要方向上取得了大量业界领先的成果。 在业务数据层面，百度透露今年"双11"期间，已有83%的开播主播使用过慧播星数字人，开播直播间数 同比增长119%，数字人带货GMV同比提升91%。目前，慧播星数字人正积极推进出海布局，已率先落 地 ...

Core Insights - Baidu has established two new departments for foundational and application model research, indicating a significant upgrade in its large model development system [1] - The foundational model department will focus on developing highly intelligent and scalable general AI models, while the application model department will optimize specialized models for specific business scenarios [1] - The leadership of these departments consists of internally cultivated talents, showcasing Baidu's commitment to nurturing young leaders and enhancing organizational vitality [1] Summary by Sections Foundational Model Development - The foundational model serves as the core carrier for technological breakthroughs, with Baidu's Wenxin large model 5.0 featuring 2.4 trillion parameters and utilizing native multimodal unified modeling technology [2] - This model integrates voice, image, and video data from the outset, distinguishing itself from industry peers that rely on later-stage integration [2] Application Model Development - Specialized models act as precise tools for translating foundational capabilities into industry applications, focusing on niche areas for value conversion [2] - These models leverage knowledge distillation technology from foundational models, enabling lighter and more efficient inference, with clear applications in finance, industry, and media [2] Systemic Support and Competitive Advantage - The synergy between the two model types is supported by Baidu's four-layer AI architecture, which includes chips, frameworks, models, and applications, providing a unique competitive advantage [3] - The Kunlun chip offers efficient computing power for stable training and inference of the 2.4 trillion parameter model, while the PaddlePaddle deep learning framework enables efficient operation of models with less than 3% active parameters [3] - Baidu's cumulative investment in AI has exceeded 100 billion yuan since the launch of Wenxin Yiyan, with plans for continued investment, enhancing resource focus, efficiency, and technological implementation [3]

11 月 26 日消息，百度新设立了基础模型研发部与应用模型研发部两大部门，标志着其在大模型领 域的研发体系迎来关键升级。其中，基础模型研发部由吴甜牵头，专注于高智能可扩展通用人工智能 大模型的研发；应用模型研发部则交由贾磊负责，聚焦业务场景所需专精模型的调优与探索。王海峰 继续担任 CTO 、 TSC 主席及百度研究院院长，新设部门均直接向百度 CEO 李彦宏汇报。这既显 示了公司对大模型战略的高度重视，也体现了百度持续推进干部年轻化。 新设立的两个大模型研发部门，构建起百度大模型 " 两条腿走路 " 的核心布局，形成分进合击的研 发新打法，而两类模型的应用价值在产业实践中已逐步凸显。 基础模型作为 AI 能力的 " 源头活水 " ，是技术突破的核心载体。百度此前发布的文心大模型 5.0 已奠定坚实基础，该模型以 2.4 万亿参数量的超大规模，采用原生全模态统一建模技术与超稀疏混 合专家架构，从训练之初便深度融合语音、图像、视频等多模态数据，实现了区别于业界后期融合方 案的原生协同能力。 此次履新的吴甜与贾磊均为百度内部培养的核心人才，吴甜 2006 年以校招生身份加入，贾磊 2011 年入职后主导语音技术 ...

11月25日，百度发布设立技术研发组织相关公告，新设基础模型研发部，负责研发高智能可扩展的 通用人工智能大模型，由吴甜负责； 新设应用模型研发部，负责业务应用场景需要的专精模型调优 和探索，由贾磊负责。王海峰继续担任CTO、TSC主席、百度研究院院长。以上部门均向百度CEO 李彦宏汇报。据了解，吴甜和贾磊都是百度培养的技术和管理人才。这体现了百度人才厚度以及公司 持续推进干部年轻化。 "两条腿"走路 早在两周前，2 0 2 5百度世界大会上，百度对外了发布原生全模态大模型文心大模型5 . 0。该 模型参数量达2 . 4万亿，采用原生全模态统一建模技术，具备全模态理解与生成能力，支持 文本、图像、音频、视频等多种信息的输入与输出。 公 开 报 道 显 示 ， 文 心 5 . 0 基 础 能 力 全 面 升 级 ， 在 多 模 态 理 解 、 指 令 遵 循 、 创 意 写 作 、 事 实 性、智能体规划与工具应用等方面表现突出，拥有强大的理解、逻辑、记忆和说服力。在 4 0 余 项 权 威 基 准 的 综 合 评 测 中 ， 其 语 言 与 多 模 态 理 解 能 力 与 Gemi n i- 2 . 5 -Pr ...

11月25日，百度发布设立技术研发组织相关公告，新设基础模型研发部，负责研发高智能可扩展的通用 人工智能大模型，由吴甜负责； 新设应用模型研发部，负责业务应用场景需要的专精模型调优和探 索，由贾磊负责。王海峰继续担任CTO、TSC主席、百度研究院院长。以上部门均向百度CEO李彦宏汇 报。据了解，吴甜和贾磊都是百度培养的技术和管理人才。这体现了百度人才厚度以及公司持续推进干 部年轻化。 年轻干部挑大梁 吴甜于2006年以校招生的身份加入百度，贾磊于2011年加入百度负责语音技术自主研发。组织架构的调 整，体现了百度持续推进人才厚度强，干部年轻化。 多项报告显示，AI技术人才在年龄结构方面呈现出显著的年轻化特点。有人力资源分析师认为，百度 是一家以技术为基因的公司，在培养高管领导力、优化组织结构上，一直以支持公司长期稳健发展为目 标。 "百度组织力成熟，高管年轻化，吴甜、贾磊均为百度自己培养出的技术+管理型人才，可以看出百度 人才厚度强，开放包容，有自己多年培养起来的年轻干将，管理组织架构健康稳定。"该分析人士分析 道。 分进合击新打法 而专精模型聚焦单一细分领域，采用高质量专属数据训练，精准掌握专业知识和复杂规 ...

市场回顾 上海证券近日发布计算机行业周报：过去一周（11.17-11.21）上证综指下跌3.90%，创业板指下跌 6.15%，沪深300指数下跌3.77%，计算机（申万）指数下跌2.74%，跑赢上证综指1.16个百分点，跑赢创 业板指3.41个百分点，跑赢沪深300指数1.03个百分点，位列全行业第7名。 以下为研究报告摘要： 英伟达发布Q3财报，数据中心业绩亮眼。11月19日，英伟达发布2026财年第三财季财务报告。财报显 示，英伟达第三财季营收为570.1亿美元，超出市场预期的549.2亿美元；数据中心Q3营收为512亿美 元，超过分析师预测的490.9亿美元，同比增长66%；净利润为319.1亿美元，同比大涨65%。英伟达预 计第四季度营收约为650亿美元，远高于分析师预测的617亿美元。黄仁勋在财报会上表示，Blackwell 产品的销量远超预期，用于云服务器的GPU产品"已经售罄"。截至2026年的未来五个季度，Blackwell与 Rubin的可见收入已经超过5000亿美元，订单量约为2000万颗GPU。 建议关注：（1）算力：寒武纪、海光信息、中科曙光、华丰科技、申菱环境、英维克、欧陆通、中恒 ...

Core Insights - Baidu has established two new research departments: the Basic Model Research Department and the Application Model Research Department, aimed at enhancing its capabilities in artificial intelligence [1][2] - The restructuring reflects Baidu's commitment to nurturing internal talent and promoting a younger management team, which is crucial for adapting to the rapid evolution of large model technologies [1][2] Group 1: Organizational Changes - The Basic Model Research Department will focus on developing high-intelligence, scalable general AI models, led by Wu Tian, while the Application Model Research Department will specialize in fine-tuning models for specific business applications, led by Jia Lei [1] - Both departments report directly to Baidu's CEO, Li Yanhong, indicating a strategic emphasis on AI development within the company's leadership [1] Group 2: Model Development Strategy - Baidu's large model strategy is characterized by a "two-pronged" approach: the Basic Model focuses on cutting-edge research and general capabilities, while the Application Model targets deep optimization for specific industries and business scenarios [2] - This dual strategy aims to accelerate the deployment of models across various industries, enhancing organizational support for AI applications [2] Group 3: Model Performance and Capabilities - The Wenxin 5.0 model, launched on November 13, features 2.4 trillion parameters and utilizes a unified modeling technique, demonstrating advanced multimodal understanding and generation capabilities across text, images, audio, and video [2][3] - In benchmark evaluations, Wenxin 5.0 has shown competitive performance, ranking second globally in text understanding and first domestically, validating the model's capabilities and potential [3] Group 4: Investment in AI - Baidu has invested over 100 billion yuan in artificial intelligence since the launch of Wenxin Yiyan in 2023, with plans to continue increasing investments in this area [3]