Core Insights - The importance of user-side data is increasing as the AI application ecosystem expands, with metrics like traffic scale, active users, and month-on-month growth becoming key indicators of product adoption and market acceptance [1] - The latest AI WEB100 rankings show that while the number of educational applications is not dominant, they are increasingly appearing across multiple dimensions, indicating a steep growth curve [1] - In January 2026, four educational AI applications were included in the statistics: Quizlet, Speak, Astra AI, and ThetaWave AI, reflecting a transition from "tool enhancement" to "learning system reconstruction" [1] Scale Perspective - Traditional learning tools remain the core representatives of educational AI applications, with Quizlet achieving 67.59 million monthly visits and 20.55 million monthly active users, indicating stable traffic and deep usage [2][3] - Quizlet's performance in both Monthly Visits and MAU rankings highlights its role as a stable entry point for users, despite not appearing in the growth rankings [4] Growth Perspective - The growth rankings reveal the most dynamic new variables in the sector, with ThetaWave AI, Speak, and Astra AI showing month-on-month growth rates of 220.91%, 112.81%, and 106.37% respectively [5][6] - ThetaWave AI's active user growth rate reached 334.83%, indicating a significant shift from trial experiences to deep retention and continuous use [7] Product Pathways and Market Directions - The four educational AI products represent different pathways and market focuses, covering various learning process stages from tool entry to system platforms and tutoring [9] - Quizlet exemplifies the reactivation path of traditional learning tools enhanced by AI, focusing on vocabulary memorization and test practice [10][12] - Speak's growth stems from a paradigm shift in language learning, emphasizing immersive practice environments through speech recognition and dialogue generation [13][15] - Astra AI aims to create a systematic learning process, transitioning from an auxiliary tool to a learning process manager, enhancing user engagement and monetization potential [16][17] - ThetaWave AI focuses on after-school tutoring and knowledge explanation, reflecting AI's accelerated penetration into core educational business areas [18][20] Overall Market Dynamics - The January 2026 AI WEB100 rankings indicate that while educational applications do not dominate in terms of total number and scale, they consistently appear in traffic and active user growth rankings, showcasing a phase of simultaneous scale consolidation and growth dynamics [21] - The presence of both stable user bases represented by Quizlet and the rapid growth of native educational applications like Speak, Astra AI, and ThetaWave AI illustrates the diverse structure of current educational AI applications [22]

（原标题：更"懂"中国的耐克，要发力了） 作者 | 谢芸子 编辑 | 张帆 进入2026年以来，耐克大中华区动作频频。 首先是高层的人事变动，现任大中华区负责人董炜将于3月31日正式卸任。同时，耐克宣布任命Cathy Sparks为新任大中华区副总裁兼总经理。 董炜于2005年加入耐克，深度参与并见证了大中华区业务发展的多个阶段，其中包括北京奥运会，以及 疫情这一"黑天鹅"事件带来的挑战。 在其任期内，耐克通过上海马拉松、耐克高中篮球联赛、ACG崇礼168等标志性活动，持续推进中国市 场的运动文化发展。 只不过，大中华区近期的业绩表现不那么"如人意"。尽管从整体的财报表现看，这家老牌体育用品集团 正缓慢复苏，逐步摆脱整个行业的下行周期、与近年来的业绩颓势。 耐克的业绩不振始于2022财年，其原因多被归结为DTC策略的受挫、产品创新节奏的变慢，以及在新锐 品牌强势崛起的同时，整个集团更多偏向潮流时尚、对体育市场失焦。 2025财年（2024年6月至2025年5月），耐克全年营收进一步下降至463亿美元，大中华区成为全球跌幅 最大市场，营收下降13%。 耐克营收表现，36氪根据财报数据制图 同年10月，已到退休之 ...

近年来，一些商标抢注者玩起了"国外商标、国内抢注"的把戏。 抢注者在亚马逊、eBay、沃尔玛等电商平台上反复搜罗，筛选出销量稳定、增长迅速的卖家，并将这些卖家的海外商标，在中国批量注册为国内商标， 然后向他们索要高额的许可费或转让费。 一些卖家的海外商标在国内被抢注之后，被索赔10万、28万不等的商标转让费，如果卖家不同意，抢注者就会在海关对卖家的货物出口进行"卡脖子"。 针对商标行业的乱象，管理机关做出了最新的回应。 2025年底，全国人大常委会通过了《商标法（修订草案）》（以下简称《修订法案》）首次审议，有望更加有效地遏制商标恶意抢注的行业乱象。 依照该"修订草案"精神，"注册后不使用、长期囤积等待套利"等行为，将被禁止。商标必须服务于真实生产和实际经营，而不能作为反复转手、投诉和索 赔的获利工具。 《修订草案》对于做抢注生意的"商标刺客"来说，无异于一记重拳，而对于在亚马逊、eBay上做生意的跨境电商卖家来说，是一个好消息。 01.商标抢注者的生意，将告终结 例如，一家公司在国内批量注册亚马逊热卖的品牌商标，覆盖服装、电子产品、化妆品等类目。这些商标所对应的卖家，大多已经在亚马逊等平台上跑出 了稳定销量 ...

2月12日晚，胖东来创始人于东来在社交平台上发布了胖东来部分战略规划。 于东来称无论将来什么时候，胖东来的最高管理层六十岁前必须退出权力岗位，保持团队的年轻化和团队活力。 此前于东来表示即将退休，他称："六十岁了，为了科学先进的发展模式，让年轻人勇敢承担使命，过年后正式退休，转为顾问，胖东来工作由胖东来决 策委员会主持。坚定为国家和社会的美好、坚定启迪更多人们轻松快乐的生活……相信一定会更加美好！" 对于此次战略规划，他再次重申： 1、胖东来永远坚定学校的性质！向社会分享追求学习研究的科学先进的生活理念和各方面的技术，尽力带动更多人轻松健康美好的生活，为国家的美好 而努力。 2、无论将来什么时候，胖东来的最高管理层六十岁前必须退出权力岗位，保持团队的年轻化和团队活力。 3、胖东来是学校性质，所以永不上市。 4、胖东来完成已规划样本目标，停止企业发展规模。 胖东来发展至今，许昌、新乡两地共有13家门店，其中5家综合型百货商场、7家中型社区超市、1家服饰鞋业类专业门店，1家中央厨房，2家物流中心， 并建立了自有品牌开发体系与线上销售平台。 2025年胖东来合计销售额超235.31亿元。其中，超市业态占据核心C位， ...

Core Insights - The article emphasizes that AI should be viewed as a tool to support existing business missions and goals, rather than a reason to change core values or business focus [2][3] - It warns against the anxiety surrounding AI, suggesting that individuals should not feel pressured to adapt their strategies solely based on AI trends [3][6] - The author advises that for those in the humanities, the best approach to learning about AI is to invest in paid courses or expert guidance rather than attempting to use AI tools directly [4][5] Group 1: Business Focus - The concept of business focus is defined as understanding what one aims to achieve in the current year, with AI serving merely as a tool to facilitate this [2] - Companies should not let AI dictate their business direction; instead, they should consider how AI can enhance their existing workflows and content [2][3] Group 2: Learning Strategies - For individuals not highly strategic, the recommendation is to pay for expertise rather than trying to navigate AI tools independently [4][5] - The article suggests that paying for professional guidance can resolve most challenges faced by those in the humanities when dealing with AI [5] Group 3: Activities During the Holiday - During the holiday season, companies should focus on content creation that resonates with their audience, leveraging the festive atmosphere to engage followers [6] - It is advised that leaders should continue their personal development and content production without succumbing to AI-related anxiety [6]

Core Insights - Anthropic has raised $30 billion in Series G funding, leading to a valuation of $380 billion, marking one of the largest private financings in tech history [1][5] - The company has achieved an annualized revenue of $14 billion within three years, demonstrating rapid growth in the AI sector [6][12] Funding Details - The funding round was led by GIC and Coatue, with participation from notable investors such as D.E. Shaw Ventures, Dragoneer, Founders Fund, and many others, including Microsoft and NVIDIA [5][11] - This financing reflects a significant investor interest in AI, despite ongoing debates about the sustainability of the AI bubble [3][11] Business Strategy - Anthropic's focus on enterprise users has been a key driver of its success, as these clients tend to have stronger payment capabilities and higher retention rates compared to consumer-focused models [9][10] - The company has not only built its user base but has also seen a tenfold increase in revenue over the past three years, indicating a robust market position [12] Product Development - The Claude Code product has emerged as a major growth engine, with annualized revenue surpassing $2.5 billion and a doubling of weekly active users since the beginning of the year [12][17] - Anthropic has expanded its offerings beyond programming to include financial analysis, data modeling, and other critical business functions, enhancing its versatility in the AI landscape [18] Market Position - The company is recognized as a leader in enterprise AI, with significant endorsements from investors and industry experts, highlighting its performance and scalability [23] - Anthropic's Claude is the only AI model that can be deployed across major cloud platforms like AWS, Google Cloud, and Microsoft Azure, showcasing its flexibility and adaptability to various business needs [23]

岚图汽车（07489.HK）日前完成港股上市的全部前置监管审批流程并更新招股书，上市确定性全面落地，预计于3月19日在香港联交所主板正式挂牌，港 股有望迎来"央国企高端新能源第一股"。 岚图汽车是2025年8月启动上市，2025年10月递交上市申请，2026年1月29日通过香港联交所聆讯，2026年2月11日获香港联交所原则性同意，历时仅四个 月，岚图汽车就完成发改委、商务部、外管局、证监会等多部门多层级审批，完成全部前置合规准备。 财报显示，岚图汽车2025年营收为349亿元，较上年同期增长80%，毛利率达到20.9%，净利更是超过10亿元。 一旦岚图汽车上市，将成为继蔚小理、零跑汽车、奇瑞汽车之后，港股市场迎来的又一新能源造车新势力。 岚图汽车的上市，是东风汽车厚积薄发的结果，其股票代码延用了母公司东风集团股份（00489.HK）的后三位数字，象征着岚图汽车对东风精神与造车 底蕴的传承。岚图也成为华中地区走出的第一家在港股上市的新能源汽车，证明了湖北在全国汽车版图举足轻重的地位，对湖北新能源汽车产业链上下游 的带动效应也会非常明显。 对于武汉经开区来说，则是在2026年开年之际，收获了一家重要IPO。 年 ...

【导读】清华大学团队推出的Dolphin模型突破了「高性能必高能耗」的瓶颈：仅用6M参数（较主流模型减半），通过离散化视觉编码和物理启发的热扩 散注意力机制，实现单次推理即可精准分离语音，速度提升6倍以上，在多项基准测试中刷新纪录，为智能助听器、手机等端侧设备部署高清语音分离开 辟新路。 视听语音分离（Audio-Visual Speech Separation, AVSS）技术旨在模拟人类的「鸡尾酒会效应」，即利用说话人的面部视觉线索（如口型变化），从背景噪 声或多人混合语音中精准提取出目标说话人的声音。这项技术在智能助听器、移动通信、增强现实及人机交互等领域具有极其重要的应用价值。 然而，长期以来，该领域面临着「性能与效率难以兼得」的困境：高性能模型往往依赖庞大的预训练参数和高昂的计算开销，难以在资源受限的边缘设备 上部署；而轻量化模型则通常以牺牲分离精度为代价，且常依赖高延迟的迭代计算。 针对这一痛点，清华大学计算机系胡晓林副教授团队提出了一种全新的高效视听语音分离模型Dolphin。 该模型通过引入离散化视觉语义表征和基于物理先验的全局-局部注意力机制，在大幅降低计算复杂度的同时，刷新了多项基准数据集 ...

Teamily AI 融资进展 Teamily AI 已经累计完成了 2000 万美元的融资，预计今年3月开启新一轮的融资计划。 产品及业务 Teamily AI 是一个以 AI 为原生核心构建的即时通讯工具，在功能上有些像"元宝派+飞书+领英"的结合。但如创始人何朝阳所说，Teamily 不想在已有的产品 上做延长线，而是要"探索AI和人类的协作，让人类更有效地连接和沟通。" 一句话介绍 Teamily AI 是一个 AI-native 即时消息应用（instant messenger），支持多个人类与多个 AI agents（代理）协作。它的核心是"代理社交网络"（agentic social network），人类与 AI agents可以在这里实时共存、互动。 产品主界面 Teamily AI 是一个社交 AI 平台，在这里，人类与 AI agent 共同存在、协作并一起进化。与目前市面上的一般对话大模型不同的是，它能够理解跨越所有群 组与频道的多模态对话，包括文本、图像、音乐、视频等等，并输出基于语境的洞察、推荐与行动方案，更好地服务于跨部门协作的场景。 跨群组记忆共享（Cross-Group Me ...

Core Insights - The article highlights the impressive performance of Gemini 3 Deep Think, which has achieved significant milestones in various benchmark tests, showcasing its advanced reasoning capabilities and potential applications in scientific and engineering fields [3][15][19]. Group 1: Performance Metrics - Gemini 3 Deep Think scored an unprecedented 84.6% on the ARC-AGI-2 benchmark, surpassing previous models that scored between 60%-70% [3][19]. - In the Humanity's Last Exam (HLE), it achieved a new state-of-the-art (SOTA) score of 48.4% [3][15]. - The model also reached a remarkable Elo score of 3455 on Codeforces, ranking it as the 8th best globally, with only 7 individuals scoring higher [1][15]. Group 2: Cost Efficiency - The upgrade of Gemini 3 Deep Think has led to an 82% reduction in reasoning costs, decreasing from $77.16 to $13.62 per task [21][15]. Group 3: Applications and Innovations - Gemini 3 Deep Think has demonstrated capabilities in analyzing sketches, modeling complex shapes, and generating files for 3D printing, indicating its utility in engineering tasks [7][15]. - The model successfully identified a subtle logical flaw in a complex mathematical paper, which had been overlooked in prior peer reviews, showcasing its potential in academic research [9][15]. - It optimized a method for growing complex crystals, achieving a precision previously unattainable, which could lead to new semiconductor materials [10][15]. Group 4: Team and Development - The development team of Gemini 3 Deep Think includes notable figures such as Yi Tay and Shunyu Yao, both of whom have significant backgrounds in AI and physics [27][28].