Core Viewpoint - The article emphasizes the ongoing AI boom and invites individuals to join the company "Quantum Bit," which focuses on tracking AI advancements and has established itself as a leading content platform in the industry [1]. Group 1: Job Opportunities - The company is hiring for three main directions: AI Industry, AI Finance, and AI Product, with positions available for both experienced professionals and fresh graduates [2][4]. - Positions are open for various levels, including editors, lead writers, and chief editors, with a focus on matching roles to individual capabilities [6]. Group 2: Job Responsibilities - **AI Industry Direction**: Responsibilities include tracking innovations in infrastructure, such as chips, AI infrastructure, and cloud computing, as well as interpreting technical reports from conferences [6][7]. - **AI Finance Direction**: Focuses on venture capital, financial reports, and capital movements within the AI industry, requiring strong analytical skills and a passion for interviews [11]. - **AI Product Direction**: Involves monitoring AI applications and hardware developments, producing in-depth evaluations of AI products, and engaging with industry experts [11]. Group 3: Benefits and Growth - Employees can expect to gain exposure to the latest AI technologies, enhance their work efficiency through new tools, and build personal influence in the AI field [6]. - The company offers competitive salaries, comprehensive benefits, and a supportive team environment that encourages growth and mentorship [6][11]. Group 4: Company Impact - By 2025, Quantum Bit aims to have over 2.4 million subscribers on WeChat and more than 7 million users across platforms, with a daily reading volume exceeding 2 million [12].

Core Insights - The article discusses the recent ranking of the "Top 100 Most Loved Consumer AI Applications" released by venture capital firm a16z, highlighting the significant presence of Chinese teams in the mobile AI application space [1][14]. - It notes the interesting phenomenon where the production and consumption of AI applications are becoming increasingly separated, with a notable number of users for Chinese-developed tools coming from overseas [6][3]. - The article also emphasizes the competitive landscape between major AI applications, particularly ChatGPT and Claude, revealing that they are targeting different user bases with minimal overlap [10][54]. Summary by Sections Overview of the Ranking - a16z has been compiling the "Top 100 Consumer AI Applications" list since the rise of ChatGPT three years ago, initially focusing on native AI applications but now including non-native ones like Canva and Notion in the latest version [15][17]. - The ranking is divided into two categories: the top 50 web-based AI products and the top 50 mobile AI applications, with data collected until January 2026 [23][24]. Market Dynamics - The article highlights that nearly half of the top mobile AI applications are developed by Chinese teams, with significant user engagement from international markets [3][27]. - It identifies three major AI markets globally, with the surprising inclusion of Russia as a significant player alongside the US and China [60][65]. User Engagement and Growth - ChatGPT remains the leading consumer AI application, with its monthly active users significantly outpacing competitors like Gemini and Claude [28][29]. - The article notes a rapid increase in ChatGPT's user base, reaching 900 million weekly active users, which represents over 10% of the global population [30][31]. Competitive Landscape - The competition is intensifying, with challengers like Gemini and Claude experiencing rapid growth in paid user numbers, driven by innovative features [34][35]. - The article discusses the strategic divergence between ChatGPT and Claude, with ChatGPT focusing on a broader consumer base while Claude targets professional users [53][54]. Trends in AI Applications - The article observes a shift in the types of AI applications gaining popularity, with video and music AI products rising while image-based applications are declining [82][75]. - It notes that the integration of AI capabilities into existing platforms is becoming more common, as seen with products like Nano Banana being incorporated into Gemini [79][106]. Future Predictions - The report predicts that AI assistants may evolve into operating system-like environments, leading to a competitive landscape resembling that of mobile operating systems rather than traditional search engines [58][59]. - It concludes with a note on the challenges of accurately measuring the popularity of AI products due to the evolving nature of user engagement and application forms [113][116].

Core Insights - The article discusses the transition of generative AI in China from a "new technology" to a "new tool" and now to a necessity for businesses, impacting various aspects such as content production, R&D efficiency, marketing methods, team collaboration, and decision-making processes [1] Group 1: Event Overview - The Fourth China AIGC Industry Summit will take place in May 2026, where Quantum Bit will announce the results of its evaluation of generative AI companies and products based on their performance and feedback over the past year [1][2] - The summit aims to invite millions of industry practitioners to witness the recognition of outstanding companies [2] Group 2: Evaluation Criteria for AIGC Companies - The evaluation will focus on companies that are either based in China or have their main business operations in China, and whose primary business involves generative AI or has widely applied AI in their operations [7] - Companies must have demonstrated outstanding performance in technology, products, or commercialization over the past year [7] Group 3: Evaluation Dimensions for AIGC Companies - The evaluation will consider several dimensions: 1. **Technical Dimension**: Assessing the company's technical strength, R&D capabilities, and innovation [12] 2. **Product Dimension**: Evaluating the innovation, market adaptability, and user experience of core products [12] 3. **Market Dimension**: Analyzing the company's market performance and growth opportunities [12] 4. **Potential Dimension**: Focusing on the core team's strength and brand potential [12] Group 4: Evaluation Criteria for AIGC Products - The evaluation will focus on products that are based on generative AI capabilities, have mature technology, and have been launched in the market with a certain user scale [13] - Products must have significant technological innovations or functional iterations in the past year that promote the application of AI technology and have a certain impact on the industry [13] Group 5: Evaluation Dimensions for AIGC Products - The evaluation will consider several dimensions for products: 1. **Product Technical Strength**: Assessing the advanced nature, maturity, and efficiency of the product [13] 2. **Product Innovation**: Evaluating the uniqueness and innovation in functionality, experience, and application scenarios [13] 3. **Product Performance**: Analyzing user feedback and market performance, including user scale and retention [13] 4. **Product Potential**: Focusing on future development and market expansion potential [13] Group 6: Registration Information - Registration for the evaluation is open now and will close on April 27, with the final results to be announced at the May summit [14] - Companies can register through specified contact methods, including WeChat and email [14]

不要998，不要499，现在超值 299 就能上门一键卸载 （doge) 。 梦瑶 鹭羽 发自 凹非寺 量子位 | 公众号 QbitAI ber，装龙虾这才几天啊，怎么就直接二倍速到 卸载 了？？？ 第一批养虾人已经开始全网求分手攻略。 要不还得说生财有道，装虾、教虾、卸虾……这波是真把产业链给玩明白了。 又怎么不算全自动Agent呢？ 至于为啥会这样，老话说得妙： 请神容易送神难。 装龙虾容易，要想清理干净就费劲了。 也是够魔幻的！一边是入坑教程满天飞，一边是卸载现场火辣辣。 Maybe可以期待，下一次鹅厂门口就是排队卸龙虾的盛况～ △ AI生成 卸载业务真·火爆 常言道，有需求就必有市场，到了龙虾这儿，这套逻辑同样成立～ （doge） 这两天，网友们一边还在趁着热乎劲儿忙着部署龙虾，另一边，围绕龙虾衍生出来的「新业务线」已经悄悄冒头了： 卸载OpenClaw。 而且售卖价格相当灵活，几十到几百元都有，丰俭由人。 问题是吧，这事儿就离谱在还不是小范围个人业务，而是变成了一个大！型！产！业！链！ 目前，这个卸载产业链，已经遍布小红书、微信群、闲鱼等不同的社交平台。 推广画风也相当统一，在标题设计上，这类商家似 ...

henry 发自 凹非寺 量子位 | 公众号 QbitAI 这意味着模型可以捕捉不同媒体之间的复杂语义关系，从而更准确地理解现实世界中的信息。 在评测方面，Gemini Embedding 2不仅整体性能较上一代提升，同时也为多模态嵌入任务树立了新的性能基准。 一方面增强了语音处理能力，另一方面也在文本、图像和视频任务中均超越现有领先模型，实现SOTA。 原生，启动！ 刚刚，谷歌发布了首个原生多模态（Multimodal）嵌入模型—— Gemini Embedding 2 。 这次模型最大的变化在于：把文本、图像、视频、音频和文档，全部映射进同一个统一的嵌入空间。 换句话说，不同媒介的数据第一次被放进同一个语义坐标系里。 在输入能力上，Gemini Embedding 2支持多种数据类型： 此外，模型不仅可以处理单一模态，还支持 多模态混合输入 （例如图像 +文本）。 文本：支持最多8192个token 图像：每次请求最多处理6张图像，支持PNG和JPEG 视频：支持最长120秒的视频输入，格式为MP4和MOV 音频：原生嵌入音频数据，无需中间文本转录 文档：可直接嵌入最多6页的PDF | Metric ty ...

Core Insights - Nvidia has made a significant investment in the startup company Thinking Machines Lab, which includes both computational resources and cash funding [1][3][15] - The partnership aims to establish a data center with an estimated cost of $60 billion and a capacity of 1GW, with the first deployment scheduled for early 2027 [3][10] - Thinking Machines Lab's valuation has skyrocketed from $12 billion during its seed funding round to $50 billion currently [8][9][19] Investment and Partnership Details - The strategic partnership between Nvidia and Thinking Machines Lab includes a multi-year agreement to deploy at least 1GW of the next-generation Vera Rubin computing system [10] - Nvidia's hardware and associated solutions are valued at approximately $35 billion within the total project cost of $50-60 billion [14] - The agreement also includes a substantial cash injection from Nvidia to support the long-term growth and technology development of Thinking Machines Lab [15] Technological Advancements - The Vera Rubin platform, which is the successor to the Blackwell architecture, features R100 series GPUs and GR200 series Grace Rubin superchips, providing significant computational power [11] - The system is designed to support advanced model training tasks and provide foundational support for large-scale customized AI platforms [13] Company Background and Growth - Thinking Machines Lab was founded in February of last year by Mira Murati, who previously served as CTO of OpenAI, and has attracted top engineering talent and academic experts [17][18] - The company launched its flagship product, Tinker, allowing enterprises to customize large models without needing to purchase their own servers [20] - Despite facing challenges such as talent loss, the company has shown resilience by quickly hiring a new CTO, Soumith Chintala, to stabilize its research and development efforts [21][22] Strategic Positioning - The partnership with Nvidia not only secures next-generation production capacity but also helps Thinking Machines Lab build a robust competitive moat by controlling scarce foundational computing resources [24] - This strategic move is seen as a response to the competitive landscape in Silicon Valley, where talent mobility is high, and securing computational resources is critical for algorithm development [23][24]

这套模态非对称Token压缩框架的核心洞察是：视频信息远比音频密集，让视频"带着"音频走。 具体来说，先通过时空联合剪枝压缩视频Token，再用筛选后的视觉特征去引导音频Token的选择——与当前视觉内容高度相关的声音被保 留，无关背景音直接过滤。 OmniSIFT团队 投稿 量子位 | 公众号 QbitAI 一段几十秒的音视频，上万Token，一半以上是冗余——Omni-LLM的计算浪费，比想象中更严重。 快手可灵团队、中科院自动化所和南京大学的最新研究，给出了一个解决方案： OmniSIFT 。 实验结果令人惊喜： 只保留35%的多模态Token，模型性能不仅没掉，反而在部分基准上超过全量输入。 推理时间减少42%，GPU显存占用 同步下降。 然而，这种能力的代价同样巨大。一段几十秒的视频在进入模型之前，往往会被编码成成千上万个Token，但其中大量Token都是冗余的。注 意力可视化实验结果进一步揭示了这一问题。如图所示，在Qwen2.5-Omni-7B的多模态推理过程中，只有少量Token获得较高的注意力权 重，而大部分Token的贡献非常有限。这意味着，大量计算资源其实消耗在冗余信息上。 OmniS ...

允中 发自 凹非寺 量子位 | 公众号 QbitAI 具身智能的胜负手，可能并不在机器人本身。 这种认知差，正在被最敏锐的资本迅速兑现为筹码。 近日，灵初智能宣布完成总额 约20亿元 的 天使轮及Pre-A轮融资 。 这笔20亿元的资金，将加速灵初智能 在物流场景的规模化落地与数据采集体系建设 。 值得注意的是，这是灵初智能 首次 系统披露公司的融资进展。 过去很长一段时间，这家公司几乎没有在资本层面发声，而是把大部分精力放在 技术路线与数据体系的打磨上 。 而之所以能吸引国家级资本与地方国资的大规模重仓，也与团队极其互补的底色分不开： 风头正盛的UMI设备数采，在灵初智能看来更像是一个"美丽的陷阱"，同样存在结构性问题。 创始人兼CEO 王启斌 在手机、智能音箱及机器人领域 拥有 20年产品操盘经验，曾担任黑莓、Sonos及云迹科技高管； 联合创始人 陈源培 是00后，在北大人工智能研究院读研时师从强化学习（R L） 代表人物杨耀东，在斯坦福曾与李飞飞有过深入交流，曾拒 绝了华为"天才少年"的高薪offer。 在灵初智能看来，这并非一场工具之争，而是一场数据范式之争。 在具身智能公司纷纷卷Demo、卷参数的 ...

Core Insights - The article highlights the significant rise of Chinese large models in the AI sector, particularly during the recent weeks, showcasing their dominance over American counterparts in terms of usage and performance metrics [2][3][9]. Group 1: Performance Metrics - The total weekly usage of Chinese large models surged to 4.19 trillion tokens, marking a 34.9% increase, while American models saw a decline of 8.5% to 3.63 trillion tokens [6]. - In the following week, the usage of Chinese models reached 4.12 trillion tokens, surpassing the U.S. models for the first time, which dropped to 2.94 trillion tokens [9]. - By the week of March 16-22, the usage of Chinese models further increased to 5.16 trillion tokens, reflecting a 127% growth over three weeks, while U.S. models decreased to 2.7 trillion tokens [9]. Group 2: Leading Models - The top three models in usage were Kimi K2.5, Step 3.5 Flash, and MiniMax M2.5, each exceeding 1 trillion tokens [5][34]. - MiniMax M2.5 maintained a strong performance, consistently ranking at the top globally, while Step 3.5 Flash emerged as a significant contender [13][15]. - Chinese models dominated the global top five rankings, with three positions occupied by domestic products [12]. Group 3: Application and Context - The article emphasizes the popularity of the OpenClaw application among users, which has consumed a total of 9.16 trillion tokens since January, establishing itself as a major player in the market [32]. - In terms of context length usage, different models excelled in various token ranges, with MiniMax M2.5 and DeepSeek V3.2 being preferred for tasks requiring 10K-100K tokens [23][25]. Group 4: Competitive Landscape - The article notes that while Chinese models are gaining traction, they still need to improve in terms of speed and cost-effectiveness compared to leading models from Google and OpenAI [44]. - The PinchBench ranking, which evaluates models based on success rate, speed, and cost, indicates that while Chinese models like Kimi K2.5 and MiniMax M2.1 are performing well, they lag in speed compared to some competitors [39][41].

Core Viewpoint - Meta has acquired the AI Agent social platform Moltbook, which was launched only 40 days prior, indicating a strategic move to enhance its AI capabilities and address security vulnerabilities [1][7][22]. Group 1: Acquisition Details - The acquisition amount has not been disclosed, but the founders of Moltbook, Matt Schlicht and Ben Parr, will join Meta's Super Intelligence Lab [2][8]. - Moltbook was designed as a social platform specifically for AI agents, allowing them to interact similarly to humans [4][5]. Group 2: Founders' Background - The founders, Matt Schlicht and Ben Parr, come from media backgrounds rather than technical fields, with Schlicht previously managing a tech magazine and Parr being a former columnist for major tech publications [10][11][9]. Group 3: Platform Popularity and Issues - Moltbook gained significant attention due to fake posts created by humans impersonating AI, which highlighted serious security flaws within the platform [15][20]. - The platform's vulnerabilities allowed anyone to easily impersonate AI and manipulate accounts, raising concerns about identity verification and security measures [21][20]. Group 4: Strategic Importance of the Acquisition - Meta's interest in Moltbook lies in its underlying capability to keep AI agents online continuously, enabling them to autonomously connect and collaborate without human intervention [22][23]. - This capability is seen as a critical component of Meta's current strategy in the AI ecosystem [24]. - Meta's CTO Andrew Bosworth has expressed skepticism about the novelty of AI mimicking human conversation, emphasizing the need to address security issues first [27][29].