Core Viewpoint - The launch of Vidu Q1's reference generation feature by Shengshu Technology represents a significant transformation in video content production, streamlining the traditional process and redefining AI-native storytelling [1][2]. Group 1: Traditional vs. AI Video Production - Traditional video production involves multiple steps including scripting, character setting, storyboarding, filming, post-production effects, editing, and final output, which can be time-consuming and requires specialized skills [1]. - The AI video production process simplifies this to scripting, image generation, video creation, editing, and final output, while Vidu Q1's reference generation feature further reduces this to just reference image input, video generation, and editing [1][2]. Group 2: Advantages of Vidu Q1's Reference Generation - The core advantage of the reference generation feature is the elimination of the storyboard phase, allowing creators to generate videos directly from reference images without the need for extensive professional skills [2]. - Vidu Q1 allows for the integration of multiple reference elements, such as characters, props, and scenes, into a single video, enhancing the efficiency and flexibility of video production [2][3]. Group 3: Flexibility and Reusability - Creators can utilize the reference generation feature to mix and match various elements, leading to numerous combinations and unique video outputs, thus increasing the reusability of materials [3]. - The feature significantly improves consistency for projects requiring high subject coherence, such as e-commerce and advertising, by supporting up to seven consistent subjects simultaneously [3].

2025.07. 05 本文字数：2033，阅读时长大约4分钟 作者 | 第一财经 吕倩 近一个月，多款视频模型新品发布，包括生数科技视频模型Vidu更新至可一键生成32秒视频，支持 音视频合成与4D生成；MiniMax发布海螺Hailuo-02，支持最高1080P、最长10秒的超清视频端到端 生成；百度（9888.HK）发布首个图生视频大模型MuseSteamer，面向包括广告商在内的专业视频内 容创作者。 但在过去几年，这一赛道并不被市场看好。 对比AI领域目前正被资本追捧的Agent（智能体），视频模型热度并不算太高。瑞银（UBS）研报认 为，视频模型领域的竞争不会像大语言模型领域的竞争那样激烈，主要是受视频模型训练所需的视频 语料内容所限。但同时，目前市面上由大厂与明星创业公司组成的战队，正在基础模型效率提高的背 景下，加快产品更新与商业化落地。 | App | Model | Monthly Standard | Monthly | Credits per | Length per | Cost (US$) | | --- | --- | --- | --- | --- | --- | --- | ...

Group 1: E-commerce and Pricing Trends - Major U.S. e-commerce platforms, including Amazon, have raised prices on nearly 1,000 products since April 9, with an average increase of about 30% due to rising operational costs linked to Trump's tariff policies and global trade tensions [2] - Cross-border e-commerce platforms like Temu and Shein have announced price adjustments starting April 25, citing increased operational costs from recent global trade rule changes [2] Group 2: Company Developments - Yushutech's founder Wang Xingxing mentioned the possibility of a future IPO in Hong Kong, where the company already has business operations and numerous collaboration opportunities [2] - Meituan has issued a statement denying rumors that riders would be permanently banned for using dual-opening or JD software, clarifying that the circulated images were manipulated [3] - JD.com plans to recruit 100,000 full-time riders over the next three months, offering benefits such as full social insurance and a guaranteed minimum salary of at least 5,000 yuan for the first three months [4] - Elon Musk's personal wealth has decreased from over 450 billion to around 300 billion in less than 100 days, primarily due to setbacks during his political engagements [4] Group 3: Technology and Innovation - WeChat has launched a new feature allowing public accounts and service accounts to conduct gift draws directly through the platform, enhancing user engagement [3] - Shengshu Technology's Vidu video model has fully integrated with Z.ai's MaaS platform, aiming to advance domestic large model technology and industry application [3][5] - Apple is reportedly advancing its smart glasses project, codenamed "N50," which will incorporate AI technology and is expected to be launched by 2027 at a more affordable price [6]

4月27日，智谱与生数科技宣布达成战略合作，将基于各自在大语言模型和多模态生成模型的技术积累 和优势，在联合研发、产品联动、解决方案整合、行业协同等多方面展开合作。根据战略协议，在产品 合作方面，智谱MaaS平台将接入生数科技Vidu API。 ...

02 "北曹"砺思资本创始合伙人、董事长曹曦 出生于1 9 8 5年的砺思资本创始合伙人、董事长曹曦，毕业于北京大学医学部，曾在腾讯公司， 金 山 软 件 就 职 ， 并 担 任 红 杉 资 本 合 伙 人 。 代 表 作 ： 月 之 暗 面 、 无 问 芯 穹 、 万 有 引 力 Gr a v it yXR 、 沐 言 智 语 、 星 迈 创 新 、 乐 享 科 技 、 剂 泰 医 药 、 逸 文 科 技 、 虚 实 之 间 Un i c h a t 、 MiAO。 进入到2 0 2 5年，全球风险投资行业全面拥抱了"AI "革命，随着1 9 8 3年出生的张一鸣，1 9 8 5年 出生的梁文锋，1 9 8 5年的山姆奥特曼，1 9 8 3年出生的达里奥阿莫迪等8 0后企业家进入全球人 工智能核心舞台。而与此同时，AI初创公司的背后也诞生了新一批由8 0后及准8 0后创始人及董 事长操盘的新一代人工智能领域投资机构也登上了历史舞台。 在中国AI大模型"六小龙"等一批新型AI公司背后的天使投资人中，也出现了聚焦于AI赛道的 新型风险投资机构的掌门创始人。包括了"南林"卓源亚洲林海卓，"北曹"砺思资本曹曦，"东 ...