图片来源： Mirage 视频编辑应用 Captions 的开发商 Mirage 已从 General Catalyst 的客户价值基金（ CVF ）筹集了 7500 万美元的增长融资。 过去一年，这家初创公司对其产品和品牌形象都进行了重大调整。 公司从 Captions 更名为 Mirage ，旨在定位为一家生产多种模型、同时服务于 广告营销等行业的人工智能实验室。 它还专门针对短视频的节奏把控、画面构图与注意力动态训练了一个模型 。 该公司还于 2025 年 1 月转向采用免费增值模式 ，以更好地与字节跳动的 CapCut 和 Meta 同年稍晚推出的 Edits 等应用竞争。 目前该公司还推出 了视频创作套件，其中包含 Captions 的部分功能，可帮助企业批量制作和分发视频。 Mirage 联合创始人兼首席执行官 Gaurav Misra 表示，公司目标是开发更多模型。但他并未具体说明下一组模型的功能，仅表示这些模型将专注于 " 组装智能 " ——即利用不同来源和组件合成视频。 参考资料： https://techcrunch.com/2026/03/24/mirage-raises-75m-to- ...

a16z 发布了全球 AI 产品 Top100 的第六版更新，相比之前的版本，有了一些明显的变化：AI 已经成为主流产品的默认功能，Gemini 追赶 ChatGPT 很凶猛，以及，真正能完成任务的通用 Agent 来了。 以下是全文的编译。 原文链接：https://a16z.com/100-gen-ai-apps-6/ 三年前，我们发布了这份榜单的第一版，目标很简单：找出哪些生成式 AI 产品真正被主流消费者使用。在当时，「AI 原生」公司和其他公司之间的界限 很清晰。ChatGPT、Midjourney 和 Character.AI 都是围绕基础模型从零构建的产品，而软件行业的其他玩家还在摸索这项技术该怎么用。 这个界限已经不成立了。CapCut 是一款拥有 7.36 亿月活移动用户的视频编辑器，其最受欢迎的功能都依赖 AI——背景移除、AI 特效、自动字幕和文生 视频。Canva 的整个增长引擎都建立在 Magic Suite AI 工具之上。Notion 的付费 AI 挂载率在一年内从 20% 飙升至超过 50%——AI 功能现在约占公司 ARR 的一半。 | | | | | | The Top 5 ...

Core Insights - Gaxos.ai Inc. has finalized a partnership with Byteplus, the Enterprise Technology Division of ByteDance, which will provide preferred pricing and significant discounts on video generation models [1][2] - This collaboration aims to enhance Gaxos' flagship AI platform, Art-Gen, by enabling faster deployment of AI capabilities, thereby improving gross margins as user adoption scales [2][3] Company Developments - The partnership positions Gaxos to penetrate the multi-billion-dollar global creator economy by integrating advanced AI video tools into its ecosystem, reducing infrastructure costs and improving scalability [3] - Gaxos anticipates expanded monetization through Art-Gen's subscription and credit-based tiers, alongside improved margins due to rebates and preferred pricing [4] Leadership Perspective - The CEO of Gaxos expressed enthusiasm about the partnership, highlighting its potential to deliver exceptional creative power to users globally while enhancing the company's economic performance [4]

Core Viewpoint - Seedance 2.0, a new video generation model by ByteDance, is rapidly gaining popularity globally, significantly lowering the barriers and costs for professional video creation [4]. Group 1: Product Features - Seedance 2.0 supports four modalities of input: image, video, audio, and text, allowing users to generate multi-shot videos with complete original soundtracks in just 60 seconds [4]. - The model has gained traction not only in China but also on overseas social media platforms, with users creating imaginative content such as "Wukong vs. Marvel heroes" and "pets dunking on NBA stars," with several demonstration videos exceeding one million views [4]. Group 2: Industry Impact - American directors and industry professionals have expressed concerns that Seedance 2.0 could disrupt Hollywood, with one director stating he created a trailer for a never-released film in just 20 minutes for $60, achieving cinematic quality [4]. - The sentiment among tech workers and AI researchers echoes this concern, with repeated assertions that "Hollywood is cooked" [19][20]. Group 3: User Accessibility - Seedance 2.0 is set to be integrated into CapCut, making advanced video creation accessible to a younger audience, specifically mentioning that every 12-year-old in America will soon have this capability on their iPhones [14][15]. Group 4: Platform Adjustments - In response to concerns about potential misuse, the platform's operators announced that during the beta testing phase, they would not support the input of real human images or videos as reference subjects, emphasizing the importance of respecting creative boundaries [26].

Group 1 - The release of ByteDance's Seedance 2.0 video model marks a pivotal moment in the global entertainment industry, indicating the widespread adoption of AI-generated content (AIGC) in short videos, micro-dramas, and future mid-to-long video platforms [1] - The enhanced model allows for the visualization of stories through one-click prompts, which will disrupt content production and challenge existing players to integrate and monetize this new wave of engaging content [1] - The micro-drama industry and IP owners are expected to benefit defensively in the near term, with companies like Reading Group (00772) poised to gain from the demand for mid-tail IPs and their own micro-drama operations and investments [1] Group 2 - Kuaishou-W (01024) has improved its K可灵 3.0 model, enhancing controllability, character consistency, visual realism, and multilingual generation capabilities, making it a strong competitor in the market [2] - The competition among generative video models is intense globally, with over 10 major companies, including Google, Vidu, xAI, OpenAI, Alibaba-W (09988), ByteDance, Kuaishou, PixVerse, Lightricks, Runway, MiniMax-WP (00100), and Luma Labs, vying for leadership [2] - Long-term, it is believed that ByteDance possesses full-stack capabilities from cloud services (Juliang Engine) to toolkits (CapCut) and distribution platforms (Douyin/Hongguo) [2]

Core Viewpoint - ByteDance officially launched the image generation model Seedream 5.0, positioning it as a practical AI creation engine aimed at disrupting the content creation industry, directly competing with Google's Nano Banana Pro [1][11]. Group 1: Technological Breakthroughs - The model's core technological advancements are evident in three areas: significant improvement in image quality, optimization of detail textures and lighting effects, enhancing the commercial usability of e-commerce posters and character portraits [3][13]. - Breakthroughs in intelligent interaction allow for real-time retrieval of raw images, addressing the traditional AI image generation's "information lag" issue, while accurately interpreting abstract commands and supporting localized brush editing [3][13]. - A deep ecological closed-loop is constructed, integrating tools like Jianying, CapCut, and Xiaoyunque, achieving full-link coverage from "generation to editing to distribution" [3][13]. Group 2: User Access and Experience - Domestic users can experience Seedream 5.0 through the image preview entry in Jianying and Xiaoyunque apps, while overseas users can access it via the CapCut integration [5][15]. - The platform offers a limited-time feature allowing 20 free generations per day, with members able to unlock unlimited generations and commercial licenses [5][15]. Group 3: Industry Impact and Future Outlook - Seedream 5.0 is expected to revolutionize creation efficiency, potentially increasing content usability for ordinary users to 90% and shortening the creation cycle by 10 times, thereby accelerating the industrialization process in the self-media and e-commerce sectors [10][18]. - The competitive logic of domestic models is shifting from "parameter competition" to "ecological integration," which may lead to increased legal risks related to copyright infringement and false content [10][18].

Core Insights - In 2025, the global mobile market reached a historic milestone as non-gaming app revenue surpassed mobile game revenue for the first time, driven primarily by generative AI [1][5][13] - The mobile market is transitioning from scale expansion to a focus on efficiency and monetization [1][5] Mobile Market Overview - The total app downloads in 2025 reached 149 billion, a year-on-year increase of 3.8%, with users spending an average of 3.6 hours daily on mobile devices [1][8] - In-app purchase (IAP) revenue grew by 10.6% year-on-year, reaching $167 billion, indicating improved monetization capabilities [1][8] Non-Gaming App Revenue - Consumer spending on non-gaming apps reached $85.6 billion in 2025, a 21% year-on-year increase, while mobile game revenue was $81.8 billion, growing only 1% [2][14] - Generative AI contributed significantly to this growth, adding $5 billion in revenue, with categories like social media, video streaming, and productivity apps also performing well [2][5] Generative AI Impact - The download volume for generative AI applications doubled to 3.8 billion in 2025, with in-app purchase revenue exceeding $5 billion, marking a nearly threefold increase [2][5] - AI assistants saw a 148% increase in downloads, while short video applications surged by 278%, becoming key drivers of non-gaming app growth [2][5] User Engagement Trends - Users averaged 34 different apps per month, with total usage time exceeding 53 trillion hours in 2025 [1][8] - Social media applications accounted for nearly 2.5 trillion hours of user engagement, reflecting their dominance in the mobile ecosystem [22][23] Market Dynamics - The mobile market is characterized by increasing competition for user attention, with gaming, social media, and AI applications vying for user time [6][19] - Local market insights remain crucial for success across various categories, influenced by regulatory environments and consumer preferences [6][19]

Core Viewpoint - ByteDance has officially launched its image generation model Seedream 5.0, which is now available in video editing applications such as Jianying, CapCut, and the AI creation platform Xiaoyunque, with a limited-time free experience on the Jimeng AI platform [1][2]. Group 1 - The new model enhances accuracy and intelligence levels, faster image creation expressiveness, and integrated online knowledge capabilities [2]. - Seedream 5.0 can deeply understand the semantics of prompts, generating images that better match user intentions with higher detail precision and clearer layouts [2]. - The model's image-to-image functionality has improved stylization effects, providing clearer details, refined textures, and balanced lighting [2]. Group 2 - The upgrade includes new editing features, allowing users to precisely select and adjust elements using brush control [2].

Core Viewpoint - TikTok has appointed Adam Presser, an internal candidate with strong ties to the entertainment industry, to lead its newly established U.S. entity, TikTok U.S. Data Security Joint Venture, amidst ongoing concerns about data security and regulatory scrutiny [1][6]. Group 1: Leadership and Structure - Adam Presser has been chosen to oversee a platform valued at least $14 billion, which has over half of its users in the U.S. [3][8]. - Presser is considered one of the most influential leaders of TikTok in the U.S. and has been viewed as a strong contender for the CEO position [3][8]. - He reports directly to Shou Zi Chew, TikTok's global leader, who in turn reports to ByteDance's co-founder and CEO Liang Rubo [3][8]. Group 2: Background and Experience - Presser has a notable educational background, having attended Harvard-Westlake School, Yale University, and further studies at Harvard Business School and Harvard Law School [3][8]. - His professional experience includes roles at Ticketmaster Entertainment in China and WarnerMedia, where he served as Executive Vice President of International Business [3][8]. Group 3: Responsibilities and Challenges - Presser's role includes overseeing the protection of U.S. user data, content moderation, and ensuring the security of the content recommendation algorithm leased from ByteDance [6][12]. - He has been actively involved in addressing content moderation issues and has responded to calls for action from influencers regarding anti-Semitism on the platform [11][12]. - The establishment of the new entity comes in response to U.S. legislative actions requiring TikTok to sever ties with ByteDance in the U.S. market to avoid potential bans [5][9]. Group 4: Political Context - Former President Donald Trump has expressed that TikTok played a significant role in his political strategy, particularly in appealing to younger voters [3][12]. - Presser's new position may facilitate closer ties with Trump, who has been a vocal supporter of the app [12].

Core Insights - The mobile market is experiencing unprecedented growth, particularly driven by AI applications, which are reshaping the landscape [5][6][54] - In 2025, global in-app purchase (IAP) revenue reached $167 billion, marking a 10.6% year-over-year increase, with a significant shift towards monetization [19][21] - AI applications saw downloads double to 3.8 billion and IAP revenue tripled to over $5 billion, indicating a robust market demand [54][56] Mobile Market Overview - The mobile application market is entering a mature phase, with stable download volumes and user engagement, while IAP revenue continues to grow rapidly [8][12] - In 2025, total downloads across iOS and Google Play increased by 0.8% to nearly 150 billion [16] - Users spent approximately 53 trillion hours on mobile applications, averaging about 3.6 hours per day per person [17] AI Application Growth - AI Assistant applications are the most popular, with significant increases in usage time, reaching 48 billion hours, a threefold increase from the previous year [58][40] - The revenue from non-gaming applications surpassed that of gaming applications for the first time, reaching approximately $85 billion, a 21% increase year-over-year [26][21] - ChatGPT generated $3.4 billion in IAP revenue, making it the fastest application to reach $3 billion in annual revenue [31][29] User Behavior and Market Dynamics - User engagement with AI applications has shifted from initial trials to regular integration into daily life, with a notable increase in interaction rates [60][57] - The competitive landscape is consolidating, with top AI publishers like OpenAI and DeepSeek capturing nearly 50% of download shares by 2025 [67][65] - The top 1% of publishers accounted for 92% of total IAP revenue, highlighting a significant concentration of market power [69][71] Emerging Trends and Future Outlook - The focus for leading AI companies is shifting towards image and video generation capabilities, which are becoming essential for user acquisition and retention [72][74] - Different generational preferences are influencing application usage, with younger users gravitating towards efficiency and creativity-focused applications [76][78] - The mobile platform is increasingly becoming the primary entry point for users engaging with AI technologies, with a growing trend of cross-device usage [61][62]