核心结论 腾讯混元开源首个工业级原生多模态生图模型，具备强语义理解与知识推理 能力。9 月 28 日，腾讯混元正式发布并开源了原生多模态生图模型，即混元 图像 3.0（HunyuanImage 3.0），参数规模 80B。这是首个开源的工业级原生 多模态生图模型。在技术架构上，它通过一个模型完成文字、图片、视频与 音频等多个模态的输入与输出，而非通过多个模型的组合实现图文理解、图 片生成等任务。混元图像 3.0 以 Hunyuan-A13B 为基础，基于 50 亿量级的图 文对，视频帧，图文交织数据和 6T 的语料数据进行了多模态生成、理解和 LLM 的混合训练，使得模型能够充分融合多任务效果，实现超强的语义理解 能力，能够响应复杂的长文本，生成长文本文字，同时具有 LLM 的世界知 识，能够利用世界知识进行推理。这意味着，混元图像 3.0 不仅拥有生图模 型的画画能力，还具备语言模型的思考能力和常识。它就像一个自带"大脑" 的画家，可以利用智能去思考图像的布局、构图、笔触，利用世界知识去推 理常识性的画面。 腾讯混元 3D 再升级。9 月 26 日，腾讯宣布混元 3D 生成模型家族添加新成 员——混元 3D ...

Group 1 - The 2025 Guangdong Province Top 100 Private Enterprises list was officially announced, with the ranking based on the 2024 revenue of the companies [1] - The entry threshold for this year's list was set at 12.332 billion yuan, indicating a strong foundation for the leading companies [1] - The revenue distribution shows a tiered structure: 4 companies exceeded 400 billion yuan, 12 companies surpassed 100 billion yuan, and 66 companies maintained revenues above 20 billion yuan [1] Group 2 - The Pearl River Delta city cluster dominates the list, with Shenzhen (40 companies), Guangzhou (22 companies), and Foshan (16 companies) contributing a total of 78 companies [1] - Dongguan and Huizhou followed with 6 and 4 companies respectively, while Zhuhai, Zhongshan, Maoming, and Zhaoqing each had 2 companies, and Jiangmen, Yangjiang, Zhanjiang, and Yunfu each had 1 company [1] Group 3 - Huawei, BYD, and Tencent ranked as the top three private enterprises in Guangdong Province for 2025, with revenues of 86.2072 billion yuan, 77.7102 billion yuan, and 66.0257 billion yuan respectively [2] - Midea and SF Express ranked fourth and fifth, with revenues of 40.9084 billion yuan and 28.4420 billion yuan respectively [2]

从地域分布看，珠三角城市群占据绝对主导地位， 深广佛三城形成稳固的"铁三角"：深圳 （40家）、广州（22家）、佛山（16家）三地合计贡献78家上榜企业 ，东莞与惠州分别以6 家和4家紧随其后，珠海、中山、茂名、肇庆各2家，江门、阳江、湛江、云浮各1家。 其中， 华为、比亚迪、腾讯位列2025广东省民营企业前三强，美的、顺丰分别位列第四和第 五 。 | CONT | 尔有大宫企业IUU弦 | | | | --- | --- | --- | --- | | 序 | 企业名称 | 地前 | 营收总额 | | 1 | 华为投资控股有限公司 | 深圳市 | 86207200 | | 2 | 比亚迪股份有限公司 | 深圳市 | 77710246 | | 3 | 腾讯控股有限公司 | 深圳市 | 66025700 | | 4 | 美的集团股份有限公司 | 佛山市 | 40908426 | | 5 | 顺丰控股股份有限公司 | 深圳市 | 28442006 | | 6 | 珠海格力电器股份有限公司 | 珠海市 | 19003807 | | 7 | TCL科技集团股份有限公司 | 惠州市 | 16496284 | | 8 | ...

9月29日讯，昨日东风汽车集团股份有限公司与深圳市腾讯计算机系统有限公司在武汉举行战略合作签 约仪式。公司董事长、党委书记杨青，腾讯集团高级执行副总裁、云与智慧产业事业群CEO汤道生等双 方代表出席签约仪式。 根据协议，双方将深度整合核心优势，依托东风汽车在研发、生产与营销领域的经验，借力腾讯在智能 化、数字化等领域的技术，共同推进智能驾驶与智能座舱创新，强化AI技术与企业数字化融合。 经观汽车 免责声明：本文观点仅代表作者本人，供参考、交流，不构成任何建议。 ...

Group 1 - The core viewpoint of the article highlights that the approval volume of game licenses in September remains high, covering various types of products and well-known manufacturers [1] - From the beginning of the year to date, a total of 1,195 domestic games have been approved, with 145 games approved in September, including titles from Tencent, miHoYo, NetEase, and others [1] - For imported games, a total of 80 games have been approved this year, with 11 games approved in September, including titles from Chuangmeng Tiandi and Xinghui Games [1] Group 2 - The game ETF (516010) tracks the animation and gaming index (930901), which selects listed companies involved in animation production, game development, platform operation, and related industry chains to reflect the overall performance of the sector [1] - The index focuses on innovation and growth in the animation and gaming field, covering the entire industry chain from content creation to distribution channels, showcasing distinct industry representation and market vitality [1]

Core Viewpoint - The article discusses the launch of Hunyuan 3D-Omni by Tencent, a unified multimodal controllable 3D generation framework that addresses the limitations of existing methods reliant on image inputs, enhancing the precision and versatility of 3D asset creation in various industries [2][5][31]. Background and Challenges - The increasing scale of 3D data has led to the rise of generative models based on native 3D representations like point clouds and voxels, with Hunyuan3D 2.1 utilizing a combination of 3D Variational Autoencoders (VAE) and Latent Diffusion Models (LDM) for efficient 3D model generation [5]. - Existing methods face challenges such as geometric inaccuracies due to single-view image inputs, difficulties in fine control over object proportions and details, and limitations in adapting to multimodal inputs [6][7]. Core Innovations of Hunyuan3D Omni - Hunyuan 3D-Omni introduces two key innovations: a lightweight unified control encoder for handling multiple control conditions and a progressive difficulty-aware training strategy to enhance robustness in multimodal integration [9][10]. - The framework supports up to four types of control signals, significantly improving the controllability and quality of generated results [9]. Key Implementation Methods - The system utilizes various control signals: 1. Skeleton for character motion control 2. Bounding Box for adjusting object proportions 3. Point Cloud for providing geometric structure prior 4. Voxel for sparse geometric hints [11][14]. Experimental Results - The model demonstrates high-quality generation of character geometries aligned with target poses when using skeleton control, showcasing its ability to maintain geometric details across various input styles [18][19]. - Bounding box control effectively adjusts object proportions, enabling intelligent geometric reconstruction, as evidenced by successful generation of complex structures [23][25]. - Point cloud inputs significantly mitigate geometric ambiguities inherent in single-view images, ensuring accurate alignment with real-world structures [25][27]. - Voxel conditions enhance the model's ability to reconstruct detailed geometric features, improving overall generation quality [27][28]. Conclusion - Hunyuan 3D-Omni represents a lightweight, multimodal, and controllable 3D generation framework that integrates various geometric and control signals without compromising the foundational model capabilities, paving the way for future advancements in multimodal 3D generation [31].

每经AI快讯，9月29日，港股科网股午后震荡走高，阿里巴巴(09988.HK)、快手(01024.HK)均涨超4%， 腾讯控股(00700.HK)涨超3%，携程集团-S(09961.HK)、美团(03690.HK)、百度(09888.HK)、网易 (09999.HK)等个股跟涨。 ...

Core Points - Major companies including Tencent, Huawei, and OPPO signed a compliance operation commitment to enhance personal information protection [1][2] - The commitment includes strict adherence to app listing review standards and responsibilities for managing personal information [1] - The initiative aims to improve privacy policies, user consent management, data processing compliance, and user rights protection [2] Group 1: Compliance Commitment - Six key application distribution platform companies committed to strict enforcement of app listing review standards and legal responsibilities [1] - The commitment emphasizes the importance of managing personal information collection and usage in compliance with the Personal Information Protection Law [1] Group 2: Privacy Policy and User Consent - Applications must provide accessible and long-term effective privacy policies, detailing personal information processing rules and user rights [2] - Operators are required to inform users about personal information processing in a clear manner, prohibiting default selections for consent [2] Group 3: Data Processing and User Rights - Operators must adhere to the principles of data minimization and necessity, avoiding blanket authorizations and frequent pop-up requests [2] - Users should be informed of their rights to access, copy, correct, delete, and restrict processing of their personal information, with a response time of 15 working days for requests [2]

数据显示，本年度百强入围门槛为123.32亿元，头部企业体量根基雄厚。营收格局呈现梯队化特征：4 家企业营收突破4000亿元大关，12家企业超1000亿元，66家企业稳居200亿元以上阵营。 南方财经记者吴佳霖广州报道 9月29日上午，在2025世界粤商大会上，2025年广东省民营企业百强榜单正式揭晓。本次榜单由广东省 工商联发布，以2024年企业营业收入为核心依据，同步发布了"制造业百强"与"服务业50强"子榜单。 | 14 | 深圳市立业集团有限公司 | 深圳市 | 8961034 | | --- | --- | --- | --- | | 15 | 广州晶东贸易有限公司 | 广州市 | 7926093 | | 16 | 心里程控股集团有限公司 | 深圳市 | 7844628 | | 17 | 玖龙纸业(控股)有限公司 | 东莞市 | 7734320 | | 18 | 深圳传音控股股份有限公司 | 深圳市 | 6871528 | | 19 | 永道控股集团股份有限公司 | 深圳市 | 6611002 | | 20 | 深圳市爱施德股份有限公司 | 深圳市 | 6582077 | | 21 | 明阳新能源投 ...

Core Insights - Tencent's Qingshi Training Camp has launched its October program, featuring ten online thematic camps focused on popular areas such as AI mini-program development and AIGC creation, catering to diverse student interests in technology [1][2] Educational Initiatives - The program allows students to create projects like AR greeting cards without prior programming knowledge, completing them in just 30 minutes, and offers official certification for completed works [2] - Different age groups are targeted with tailored designs, including AI-generated storytelling for middle schoolers and specialized courses for college students, providing real-world experience through Tencent's business scenarios [2][10] Community Engagement - Tencent emphasizes improving the educational ecosystem by hosting events like "Qingshi Open Day," where educators can engage with technology experts and students can participate in hands-on projects with engineers [3] Social Impact - The training camp addresses the "double reduction" policy by providing enriching non-academic growth options, focusing on skills like AIGC creation and AI mini-program development, which align with the Ministry of Education's standards [4] - The program includes initiatives for underprivileged areas, such as the "AI Cultural Creation Spring Bud Class," which introduces AI tools to students from less developed regions, fostering interest in technology [6] Technological Integration - The camp promotes a collaborative experience where students use technology to solve real-world problems, emphasizing that the ultimate goal of technology is to serve as a bridge rather than a barrier [9] Program Expansion - Tencent's Qingshi Training Camp has attracted over 600,000 youth participants and collaborated with over 2,000 schools, offering 438 camp classes and nurturing numerous tech projects [10] - The program will continue to evolve, expanding its "4 + 2" layered training system to better meet the needs of students at different growth stages [11]