AI 技术迭代赋能主业成长，可灵功能创新或加速商业化变现线上营销方面，公司迭代线上营销投放产 品，升级端到端生成式推荐大模型OneRec，首创生成式强化学习出价模型G4RL，2025Q3AI 带动国内 线上营销服务收入约4-5%的提升，AIGC 营销素材带来的服务消耗额超30 亿元，UAX（全自动投放） 在外循环的消耗渗透率超70%。电商方面，端到端生成式检索架构OneSearch 实现更精准的商品匹配并 优化用户体验，驱动商城搜索订单量提升近5%，OneRec 为电商商城信息流带来GMV 高单位数提升。 直播方面，可灵AI支持AI 万象礼物定制功能上线，上线当日用户付费制作并送出礼物超10 万次。 公司持续提升基础模型能力，可灵AI 2.5 Turbo 模型发布10 日后登顶ArtificialAnalysis 文/图生视频榜 单；Q3 先后推出灵动画布，升级首尾帧功能并上线数字人解决方案，2025Q3 可灵AI 收入超3 亿元，功 能创新或加速商业化变现。 内容生态日趋繁荣，丰富商品供给及加强用户转化或驱动电商延续高增Q3 用户规模及深度双增长，平 均DAU/MAU 创新高，分别达4.16/7.31 亿（ ...

机构：东方证券 研究员：项雯倩/李雨琪/金沐阳 核心观点：主业稳健，下半年AI 赋能广告、电商核心主业提速，对后续增长形成良好支撑，我们对应 25-27 年上调收入预测，目标价102.81 港元，维持"买入"评级。 Q3 电商收入好于预期，Q4 达人抽佣相对温和、佣金收入增速环比放缓。我们测算Q3 电商收入同比增 长21%（好于此前16%预期），主要系达人分销、营销托管和大链接等商家增值服务，带来电商佣金收 入同增32%。内循环广告Q3 同增15%，与GMV 增速基本一致。25 年公司电商策略上注重流量补贴+抽 佣增加打法，Q4 抽佣相对温和，且因双11 大促品类结构mix 影响，佣金TR 处于年内相对低位，我们 预计Q4 电商GMV 同增13%，佣金/内循环广告分别同增17%/13%。 国内广告增长稳健，AI 赋能支撑中长期广告增速。我们测算国内广告Q1-Q4 yoy 分别为 可灵Q3 收入超3 亿，全年指引上调至1.4 亿美元。根据全年指引我们测算可灵Q4 收入环比相对稳定， 我们判断主要是降价导致，9 月可灵2.5turbo 降价30%，产品力+短期降价有利于在竞争初期保持市场份 额，判断付费用户数涨势 ...

Core Insights - The article introduces a novel AI model called VANS, which generates videos as answers instead of traditional text responses, aiming to bridge the gap between understanding and execution in tasks [3][4][5]. Group 1: Concept and Motivation - The motivation behind this research is to utilize video, which inherently conveys dynamic physical world information that language struggles to describe accurately [5]. - The traditional approach to "next event prediction" has primarily focused on text-based answers, whereas VANS proposes a new task paradigm where the model generates a video as the response [8][9]. Group 2: Model Structure and Functionality - VANS consists of a visual language model (VLM) and a video diffusion model (VDM), optimized through a joint strategy called Joint-GRPO, which enhances collaboration between the two models [19][24]. - The workflow involves two main steps: perception and reasoning, where the input video is encoded and analyzed, followed by conditional generation, where the model creates a video based on the generated text title and visual features [20]. Group 3: Optimization Process - The optimization process is divided into two phases: first, enhancing the VLM to produce titles that are visually representable, and second, refining the VDM to ensure the generated video aligns semantically with the title and context of the input video [25][28]. - Joint-GRPO acts as a director, ensuring that both the "thinker" (VLM) and the "artist" (VDM) work in harmony, improving their outputs through mutual feedback [34][36]. Group 4: Applications and Impact - VANS has two significant applications: procedural teaching, where it can provide customized instructional videos based on user input, and multi-future prediction, allowing for creative exploration of various hypothetical scenarios [37][41]. - The model has shown superior performance in benchmarks, significantly outperforming existing models in metrics such as ROUGE-L and CLIP-T, indicating its effectiveness in both semantic fidelity and video quality [46][47]. Group 5: Experimental Results - Comprehensive evaluations demonstrate that VANS excels in procedural teaching and future prediction tasks, achieving nearly three times the performance improvement in event prediction accuracy compared to the best existing models [44][46]. - Qualitative results highlight VANS's ability to accurately visualize fine-grained actions, showcasing its advanced semantic understanding and visual generation capabilities [50][53]. Conclusion - The research on Video-as-Answer represents a significant advancement in video generation technology, moving beyond entertainment to practical applications, enabling a more intuitive interaction with machines and knowledge [55][56].

Group 1: Market Overview - On November 21, the Hong Kong stock market saw a net inflow of 105 million HKD from northbound trading, with a net buy of 498 million HKD from the Shanghai Stock Connect and a net sell of 393 million HKD from the Shenzhen Stock Connect [2] - The most bought stocks included Tencent (00700), Xiaomi Group-W (01810), and Alibaba-W (09988), while the most sold stocks were the Tracker Fund of Hong Kong (02800), Hua Hong Semiconductor (01347), and Ganfeng Lithium (01772) [2] Group 2: Stock Performance - Alibaba-W had a net inflow of 8.47 billion HKD, with a buy amount of 41.80 billion HKD and a sell amount of 33.33 billion HKD [3] - Xiaomi Group-W recorded a net inflow of 9.48 billion HKD, with a buy amount of 29.79 billion HKD and a sell amount of 20.31 billion HKD [3] - Tencent Holdings had a net inflow of 9.61 billion HKD, with a buy amount of 22.87 billion HKD and a sell amount of 13.25 billion HKD [3] Group 3: Sector Insights - Northbound funds are actively buying technology stocks, with Tencent, Alibaba, Kuaishou-W, and Meituan-W receiving significant net buys [6] - Xiaomi Group-W's strong third-quarter performance, with a revenue increase of 22.3% to 113.1 billion HKD and a net profit growth of 80.9% to 11.3 billion HKD, is driven by its high-end smartphone strategy and automotive business [6] - Semiconductor stocks like SMIC (00981) and Hua Hong Semiconductor (01347) faced net sells of 1.89 billion HKD and 3.37 billion HKD respectively, amid reports of potential delays in U.S. semiconductor import tariffs [7] Group 4: Commodity Market Impact - Ganfeng Lithium (01772) experienced a net sell of 1.9 billion HKD, influenced by recent adjustments in lithium carbonate futures trading fees and limits, leading to a drop in futures prices [7] - The overall sentiment in the lithium market is shifting as recent price movements are increasingly driven by market speculation rather than fundamental supply-demand dynamics [7] Group 5: ETF Performance - The Southern Hang Seng Technology ETF (03033) saw a net buy of 72.34 million HKD, while the Tracker Fund of Hong Kong (02800) faced a significant net sell of 5.142 billion HKD [8] - The divergence in ETF performance reflects broader market uncertainties, particularly regarding U.S. interest rate policies and inflation concerns [8]

Core Viewpoint - Kuaishou-W has been actively repurchasing its shares, indicating a strategy to enhance shareholder value amidst market fluctuations [1] Summary by Category Share Buyback Activity - On November 21, Kuaishou-W repurchased 750,000 shares at prices ranging from HKD 63.050 to HKD 64.250, totaling HKD 47.77 million [1] - The stock closed at HKD 64.000 on the same day, reflecting a decline of 1.46%, with a total trading volume of HKD 24.52 billion [1] - Since November 20, the company has conducted buybacks for two consecutive days, acquiring a total of 1.65 million shares for a cumulative amount of HKD 106 million [1] - Year-to-date, Kuaishou-W has executed 28 buybacks, acquiring a total of 43.89 million shares for a total expenditure of HKD 2.278 billion [1] Detailed Buyback Information - The buyback details include various dates, share quantities, highest and lowest prices, and total amounts spent, showcasing a consistent buyback strategy throughout the year [2]

Core Points - Kuaishou-W (01024) announced a share buyback of 750,000 shares for a total cost of HKD 47.77 million on November 21, 2025 [1] - The buyback price per share ranged from HKD 63.05 to HKD 64.25 [1] - Additionally, the company issued 15,000 shares under the post-IPO stock option plan and 618 shares under the pre-IPO employee stock ownership plan on the same date [1]

Core Viewpoint - Kuaishou's Q3 2025 financial report shows steady revenue and profit growth, indicating a shift from expansion to value creation in its business strategy [1][3]. Financial Performance - Kuaishou reported Q3 revenue of 35.6 billion RMB, a year-on-year increase of 14.2% [1]. - Adjusted net profit reached 4.986 billion RMB, up 26.3% year-on-year [1]. User Data - The average daily active users (DAU) of Kuaishou reached 416 million, a 2.1% increase year-on-year [2]. - The average monthly active users (MAU) stood at 731 million, reflecting a 2.4% year-on-year growth [2]. Strategic Shift - Kuaishou is transitioning from a horizontal growth narrative to a focus on vertical value creation, as indicated by its stable double-digit growth in revenue and profit despite a slow increase in user numbers [3]. - The company has restructured its local services division, integrating it with the advertising business to enhance resource allocation and ensure more predictable returns [5]. AI Integration - AI is a central theme in Kuaishou's operations, with significant contributions to revenue from online marketing services, which generated 20.1 billion RMB in Q3, a 14% increase [6]. - The introduction of generative AI models has contributed approximately 4%-5% of revenue growth in the domestic market [6]. - Kuaishou's R&D expenditure reached 3.65 billion RMB in Q3, a 17.7% increase, primarily driven by investments in AI technology and talent [9]. AI Product Development - Kuaishou's Keling AI generated over 300 million RMB in revenue in Q3, highlighting its strategic importance despite its small financial weight in the overall revenue [7]. - The company is focusing on AI-driven content creation, particularly in the realm of "manhua" (comic) production, leveraging AI to enhance efficiency and reduce costs [16][18]. Market Competition - The competitive landscape is intensifying with the emergence of new AI-driven platforms, prompting Kuaishou to adapt its strategy rather than confront competitors head-on [10][11]. - Kuaishou's approach to AI differs from competitors like Sora, focusing on professional content creators first before expanding to consumer applications [12]. Future Outlook - The potential for AI in content production is significant, particularly in the "manhua" sector, which is experiencing rapid growth due to AI's efficiency [21]. - However, the market for AI-generated content may face challenges in consumer adoption, as the target demographic is already saturated with entertainment options [22].

Core Viewpoint - Kuaishou Technology (01024.HK) announced a share buyback of 750,000 shares at a cost of HKD 47.77 million on November 21 [1] Group 1 - The company executed a buyback program, indicating confidence in its stock value [1] - The total expenditure for the buyback was HKD 47.77 million, reflecting a strategic move to enhance shareholder value [1] - The number of shares repurchased was 750,000, which may impact the overall share supply and potentially influence stock price [1]

FF305 翌日披露報表 (股份發行人 ── 已發行股份或庫存股份變動、股份購回及/或在場内出售庫存股份) 表格類別： 股票 狀態： 新提交 公司名稱： 快手科技（於開曼群島註冊成立以不同投票權控制的有限公司） 呈交日期： 2025年11月21日 | 第一章節 | | | | | | | | | | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1. 股份分類 | | 不同投票權架構公司普通股 | 股份類別 | B | | 於香港聯交所上市 | 是 | | | | 證券代號 (如上市) | 01024 | | 說明 | | | | | | | | 多櫃檯證券代號 | 81024 | | RMB 說明 | | | | | | | | A. 已發行股份或庫存股份變動 | | | | | | | | | | | 事件 | | | | 已發行股份（不包括庫存股份）變動 | | 庫存股份變動 | 每股發行/出售價 (註4) | | 已發行股份總數 | | | | | 已發行股份（不包括庫存股份）數 目 | | 佔有關事件前的現有已發 行股 ...

Core Insights - On November 21, 2025, Alibaba-W (09988), Xiaomi Group-W (01810), and Yingfu Fund (02800) were the top three stocks by trading volume in the Southbound Stock Connect, with trading amounts of 75.13 billion, 50.10 billion, and 39.92 billion respectively [1] - In the Shenzhen-Hong Kong Stock Connect, Alibaba-W (09988), Xiaomi Group-W (01810), and Tencent Holdings (00700) also ranked as the top three, with trading amounts of 59.52 billion, 26.55 billion, and 24.41 billion respectively [1] Southbound Stock Connect - Top Active Companies - Alibaba-W (09988) had a trading amount of 75.13 billion with a net buying amount of +8.47 billion [1] - Xiaomi Group-W (01810) recorded a trading amount of 50.10 billion with a net buying amount of +9.48 billion [1] - Yingfu Fund (02800) had a trading amount of 39.92 billion but a net selling amount of -37.55 billion [1] - Tencent Holdings (00700) saw a trading amount of 36.12 billion with a net buying amount of +9.61 billion [1] - SMIC (00981) had a trading amount of 31.12 billion with a net buying amount of +1.49 billion [1] - Meituan-W (03690) recorded a trading amount of 12.35 billion with a net selling amount of -55.12 million [1] - Ganfeng Lithium (01772) had a trading amount of 11.02 billion with a net selling amount of -1.91 billion [1] - Pop Mart (09992) recorded a trading amount of 10.97 billion with a net selling amount of -6.97 million [1] - Hua Hong Semiconductor (01347) had a trading amount of 10.81 billion with a net selling amount of -6.86 million [1] - China National Offshore Oil Corporation (00883) recorded a trading amount of 9.73 billion with a net buying amount of +650.39 million [1] Shenzhen-Hong Kong Stock Connect - Top Active Companies - Alibaba-W (09988) had a trading amount of 59.52 billion with a net buying amount of +3.10 billion [1] - Xiaomi Group-W (01810) recorded a trading amount of 26.55 billion with a net buying amount of +3.20 billion [1] - Tencent Holdings (00700) saw a trading amount of 24.41 billion with a net buying amount of +7.75 billion [1] - SMIC (00981) had a trading amount of 24.07 billion with a net selling amount of -3.39 billion [1] - Southern Hang Seng Technology (03033) recorded a trading amount of 14.79 billion with a net buying amount of +723.43 million [1] - Yingfu Fund (02800) had a trading amount of 14.01 billion with a net selling amount of -13.88 billion [1] - Meituan-W (03690) recorded a trading amount of 10.17 billion with a net buying amount of +2.01 billion [1] - Hua Hong Semiconductor (01347) had a trading amount of 8.68 billion with a net selling amount of -2.69 billion [1] - XPeng Motors-W (09868) recorded a trading amount of 7.82 billion with a net buying amount of +1.34 billion [1] - Kuaishou-W (01024) had a trading amount of 6.63 billion with a net buying amount of +2.36 billion [1]