思宇年度活动回顾： 首届全球眼科大会 | 首届全球骨科大会 | 首届全球心血管大会 | 首届全球医美科技大会 即将召开： 2025年7月17日，第二届全球医疗科技大会 2025年9月3-5日，第三届全球手术机器人大会 2025年6月17日， 华大智造掌上无线彩色多普勒超声诊断仪EF6系列 （型号包括EF6-CLA、EF6-CLD、EF6- CLG、EF6-CLP、EF6-CLS）正式获得江苏省药品监督管理局颁发的医疗器械注册证 （注册证编号：苏械注准 20252061068） 。 该注册证的颁发标志着国内首款 64通道双探头掌上超声诊断设备 完成国家级安全性与有效性验证，正式取得 合法上市资质。 作为便携超声领域的一项关键进展，EF6系列的注册通过，不仅代表着技术参数和应用能力的全面升级，也标 志着中国便携超声设备在产品形态、图像质量与临床适配性方面，开始迈入"高清多模态"的阶段。 这是继远程超声机器人MGIUS-R3、H1系列掌上超声之后，华大智造在超声产品线中的又一重要技术成果，进 一步丰富了其"智能+远程+自动化"医疗影像生态系统。 # 产品机制与设计理念 EF6系列定位为新一代掌上超声旗舰机型，在结构 ...

Summary of Conference Call Records Industry Overview - The gaming sector is experiencing stable policies focusing on three main themes: overseas expansion, technology, and new consumption. Starting from September 2024, there will be an emphasis on promoting high-quality games for cultural export, with consumption policies for gaming and entertainment set to boost in March 2025 [5][6][7]. - Regional policies in Guangdong and Zhejiang are being implemented to support the gaming industry, with Guangdong focusing on industrial chain collaboration and Zhejiang emphasizing international ecological development [5][6]. Key Points on Gaming Sector - The approval process for game licenses is normal, with a slight increase in the number of licenses issued monthly for both imported and domestic games [5][6]. - From May 2025, there has been a noticeable increase in the supply of gaming products, particularly benefiting mid-sized developers and mobile games, as major players like Tencent and NetEase focus more on PC games [6][7]. - A-share leading gaming companies are currently valued at around 20 times earnings, with potential to rise to 25 times as product diversity increases. Recommended companies include Gigabit, Giant Network, Kaiying Network, and Hong Kong-listed Xindong [7]. AI Video Industry Insights - The AI video sector is seeing significant advancements in multi-modal technology, with Kuaishou being a standout performer. The company is expected to achieve a valuation premium due to its strong commercialization capabilities [8][9]. - Kuaishou is utilizing a dual-driven monetization model, focusing on both paid penetration and scenario expansion to achieve profitability. The company is exploring various fields including film production, advertising, and game development [8]. Financial Projections for Kuaishou - Kuaishou's AI video tools are estimated to be valued at approximately $6 billion, with projected revenue of $200 million by the end of 2025. The expected net profit for Kuaishou in 2025 is around 20.1 billion yuan, based on a 30 times valuation multiple [9]. Film Industry Performance - The film industry has seen a significant decline in box office revenue, with May's total box office at approximately 1.742 billion yuan, down 41% year-on-year. The number of viewers also dropped by 40% [10]. - The upcoming summer film season is expected to show greater elasticity and recovery, with a larger capacity for headlining commercial films. The summer season runs from June 1 to August 31, with a mix of imported and domestic films scheduled for release [10][11]. Digital Media Performance - In May, active users for major digital media platforms were reported as follows: iQIYI (350 million), Tencent Video (370 million), Mango TV (280 million), and Youku (200 million). Mango TV and Youku saw increases, while iQIYI and Tencent Video experienced declines [12][13]. - The summer period is considered the best window for historical dramas and major S-level series, with several anticipated releases already in the pipeline [13]. Conclusion - The gaming and AI video industries are poised for growth, driven by favorable policies and technological advancements. The film industry is expected to recover during the summer season, while digital media platforms continue to adapt to changing viewer preferences.

Core Viewpoint - The AI era presents a significant shift from the mobile internet paradigm, emphasizing "implementation" over mere "connection," leading to unprecedented opportunities for entrepreneurs in the AI space [1][5][6]. Group 1: Old vs New Paradigm - The old mobile internet paradigm focused on connecting large user bases and applications, while the new AI paradigm emphasizes depth and high-value implementation [4][6]. - Major tech companies are still operating under the old paradigm, which creates space for new entrants to focus on specific, high-value applications that these giants cannot fully address [5][6]. Group 2: Model Dividend - The current model dividend represents the largest opportunity in history, driven by rapid advancements in AI models since late last year [10][11]. - Companies leveraging new model capabilities in niche markets have seen significant success, with some achieving valuations exceeding $5 billion [12][15]. - The speed of achieving revenue milestones in AI has accelerated, with companies reaching $1 million in annual revenue much faster than in previous tech waves [7][11]. Group 3: Opportunities in Agent and Multimodal - The next major opportunities lie in the development of Agent capabilities and multimodal applications, which are expected to see rapid advancements in the coming year [30][31]. - The ability of models to perform complex tasks and integrate various tools is still in its early stages, indicating a significant growth potential [33][34]. - The B2B sector remains underexplored for multimodal applications, presenting a substantial opportunity for innovation [35][36]. Group 4: Market Dynamics - Entrepreneurs should focus on high-value, specific problems rather than large-scale user acquisition, as the model capabilities allow for significant impact with smaller user bases [18][19]. - The global market presents vast opportunities, and companies should not limit themselves to domestic markets but rather seek to address pain points across various industries worldwide [21][22]. - Successful companies are those that can identify and solve specific industry challenges using advanced AI models, leading to substantial competitive advantages [23][24].

Core Insights - YunJinWei, a company focused on developing embodied intelligent operating systems, recently completed a Series A+ funding round, raising 10 million yuan to enhance its platform, expand product offerings, and increase ecological coverage in various industry scenarios [1][3] - The global market for embodied intelligent devices is projected to exceed $25 billion by 2024, with a compound annual growth rate (CAGR) of nearly 20%, and China's demand for intelligent transformation in industrial automation and smart cities accounts for over 35% [1][2] - The company aims to address the urgent need for multimodal AI in physical environments, as traditional language models can only handle one-dimensional text data, while industries require integration of visual, sensor, and control command data [1][2] Technology and Innovation - YunJinWei's proprietary YunJin OS utilizes the MaM (Model-Alloy-Model) synthesis model, which achieves nanosecond-level collaborative scheduling of heterogeneous models, significantly improving efficiency in scenarios like intelligent inspection [2] - The architecture addresses the challenge of fragmented physical world data by allowing over 90% of private multimodal data to be processed on edge devices, thus reducing data security costs [2] - The VT-Transformer framework developed by YunJinWei reduces model inference latency to 12ms and decreases memory usage by 85%, enabling billion-parameter multimodal models to run on cost-effective edge hardware [2] Market Penetration and Vision - As of Q2 2025, YunJinWei has served over 120 enterprises, generating revenue in the tens of millions, with notable clients including China Electronics, Guiyang Rail Transit, SAIC Group, and Shanghai Tunnel [3] - The founder, Wang Wenyi, emphasizes the vision of making AI accessible to every enterprise, facilitating low-cost training and inference for intelligent systems [3] - The team comprises experienced professionals from various fields, including system software, chip design, and visual AI, and has established partnerships with research institutions to enhance its technological capabilities [3]

从2月底开始，锦秋基金决定开始一个固定节目——每周五晚上，我们在不同城市组织一场小饭桌，把AI创业者们聚在一起好好吃顿饭。 没想到，这个"不正经的正经事"越办越有意思。 每期的人员构成"越来越MOE"——从技术极客到产品大牛，从初创founder到上市公司高管，从技术专家到独立开发者； 话题也越来越"多模态"——从芯片架构聊到出海策略，从多模态技术聊到用户心理； 甚至形式都在进化——从饭桌拓展到了茶桌。 在这里，可以暂时放下BP和估值，跟一群同样疯狂的人边吃边聊聊那些"还不太成熟"的想法。 对于刚知道锦秋小饭桌的朋友，简单介绍一下：锦秋小饭桌是一个每周五晚在北京、深圳、上海、杭州等地举办的AI创业者闭门社交活动。我们把最前沿的创业 者、投资人、技术大牛聚在一起，围着一桌好菜，聊那些在办公室里不会聊的真话： 不是路演，是真·吃饭 ：没有PPT轰炸，只有一桌好菜和实打实的干货分享 不仅是networking，更是brainstorming ：深度探讨技术趋势、产品机会、商业洞察 从2月26日的第一顿晚餐，到现在已经开了 15场小饭桌 ，覆盖 北京、深圳、上海、杭州4个城市 。 在正式开始笔记之前，先预告一下近期活 ...

Core Viewpoint - The development of embodied intelligent large models is transforming traditional serving robots into more coach-like entities, but creating a true AI coach remains a long-term market challenge [1] Group 1: Market Dynamics - The cost of using serving robots is significantly lower than that of human coaches, with prices for robot sessions around 80 yuan per hour compared to 150 yuan for human coaches [2] - The current serving robots lack sufficient intelligence, primarily offering basic parameter settings without advanced features like strategy generation and feedback adjustment [2][4] - The market for serving robots is expanding, with a notable increase in consumer orders, which now exceed 50% of total orders, indicating a shift towards broader customer bases beyond professional athletes [6] Group 2: Technological Challenges - Most serving robots still utilize a modular architecture rather than an end-to-end model, which complicates real-time data processing necessary for quick responses in table tennis [4] - Developing a more generalized "sports ChatGPT" requires overcoming complex engineering challenges, including integrating image, action, and language data to create effective training strategies [6][7] - The industry is expected to see increased investment in research and market education to enhance the models' generalization and fault tolerance capabilities, which are crucial for commercial success [7] Group 3: Future Opportunities - The global market for tennis serving machines is projected to grow from $27.4 million in 2024 to $40.3 million by 2035, indicating potential for expansion in related sports technology [6] - Recent funding rounds for companies like Chuangyi Technology suggest a positive outlook for investment in the serving robot sector, highlighting the industry's growth potential [6]

Core Viewpoint - The article discusses the performance of Xiaomi's MiMo-VL-7B multi-modal model, highlighting its strengths and weaknesses compared to the Qwen2.5-VL model, particularly in various testing scenarios. Group 1 - MiMo-VL-7B model outperforms several multi-modal understanding models, especially Qwen2.5-VL, in various tests [3][5]. - The testing results indicate that the SFT (Supervised Fine-Tuning) and RL (Reinforcement Learning) versions of MiMo-VL-7B show similar performance, while the "think" version significantly outperforms the "no-think" version [5][6]. - MiMo-VL-7B's performance in recognizing handwritten OCR is noted to be poor [5][9]. Group 2 - In table recognition tasks, MiMo-VL-7B's "think" model performs well, while the "no-think" model and Qwen2.5-VL struggle [9][10]. - For medium complexity tables, MiMo-VL-7B-SFT "think" model approaches correctness, while other models fail [18][19]. - The article emphasizes that MiMo-VL-7B-SFT "think" model shows better results in complex table recognition compared to its counterparts [26][27]. Group 3 - The article concludes that Xiaomi's MiMo-VL model is impressive overall, particularly the "think" model, which excels in most capabilities except for handwritten OCR [67][68]. - Despite its strengths, the article suggests that the claims of MiMo-VL-7B significantly outperforming the 72B model may be exaggerated [68].

下一步，北京市植物保护站联合相关科研机构将在持续提升系统智能识别、智能预警等服务能力的同 时，研发优化病虫害智能巡检机器人等智慧植保硬件设备，为未来的无人化监测预警与防控作业提供智 慧植保硬件设备支撑，并通过智慧植保硬件设备筛选评价与整合平台建设，共同打造出"软件服务系统 研发+硬件设备筛选评价+软硬件融合示范推广"于一体的北京智慧植保新名片，为进一步提升我国智慧 植保的国际影响力继续贡献北京力量。 据悉，依托国内AI大模型等技术的快速发展，北京智慧植保服务系统在今年实现了两个重要突破：一 是系统新增了小麦、玉米、大桃等作物，服务覆盖作物种类增加到53种，整体服务覆盖病虫种类增加到 711种，病虫害智能识别种类增加到347种（其中蔬菜病虫230种），并且新增了AI全语音智能问答等功 能，满足了众多眼花、书写不便用户的使用需求，系统整体服务能力与使用便捷性都得到了极大提升； 二是"神农植保多模态大模型1.0"成功研发并开放使用。该模型由北京市植物保护站与中国农业大学神 农大模型研究团队联合研发，在原神农大模型基础上，新增了5万余条病虫防控技术信息、40万条高质 量标注的植保图像数据和3万条高质量植保问答数据，成 ...

Group 1 - The core viewpoint of the articles is that iFlytek is actively addressing investor concerns regarding its products and services, particularly the Robot Super Brain platform and the Spark Model [1][2] - iFlytek's Robot Super Brain platform utilizes a combination of audiovisual integration and advanced large model technology, offering a new interactive experience through a hardware-software integrated approach. The charging model includes both per-unit licensing and customized service fees [1] - Investors have suggested that iFlytek should provide full recordings of executive speeches and participation in various events on platforms like Weibo, Bilibili, and Douyin to keep small shareholders informed. The company expressed its commitment to optimizing communication methods while adhering to partner rules and compliance [1] Group 2 - Investors have high expectations for iFlytek's Spark Model, noting that it still lags behind GPT-3 in multimodal capabilities, particularly in complex image recognition tasks. Enhancements in these areas could lead to more personalized learning experiences [2] - iFlytek's management has committed to continuously improving the multimodal capabilities of the Spark Model by integrating algorithms, data, and application scenarios, with plans to promote the fusion of technology and application based on development progress [2]

作者丨 量子位 编辑丨 量子位 点击下方 卡片 ，关注" 具身智能之心 "公众号 >> 点击进入→ 具身 智能之心 技术交流群 更多干货，欢迎加入国内首个具身智能全栈学习社区 ： 具身智能之心知识星球 (戳我) ， 这里包含所有你想要的。 CVPR 2025落下帷幕，这次关注度和社交参与感，非常深度了。 比如随手抓住一只何恺明，直接变成追星现场。 在以谷歌/Meta等国际巨头为主导的展区里，中国企业规模创纪录，像腾讯、字节等大展区里面人从众。 展台面前排队体验的技术Demo，妥妥都是技术风向标~ 每一年被CVPR接收的论文大家都会关注，因为它们一定代表着最最前沿的技术风向。尤其是获得了最佳论文奖项的成果，那就得好好拜读一 下。 如果你的论文能被CVPR接收，相当于受到非常大的认可。因此相关从业者一有机会都想冲一波，万一就被录用了呢。 总结下来，有这样几个有意思的发现。 首先， 多模态、3D生成 是此次论文接收和现场研讨的热门方向，尤其像3D生成是亮点，背后高斯泼溅技术成为此次论文标题出现次数最多 的前五关键词之一。 其次， 对于基础模型的讨论远比以往更加深入，并且延伸到了产业落地 。具身智能、机器人AI在Wo ...