Core Insights - The embodied intelligence industry is currently in an acceleration phase, integrating various cutting-edge disciplines such as artificial intelligence, mechanical engineering, sensors, new materials, and brain science, with open-source ecosystems being crucial for innovation [1][6] - The Beijing Humanoid Robot Innovation Center focuses on key technology breakthroughs and industry ecosystem development, utilizing the open-source general robot platform "Tiangong" and the large-scale multi-configuration intelligent robot dataset and benchmark "RoboMIND" [1][2] Group 1: Open-source Initiatives - The "Tiangong" platform is the world's first full-size electric humanoid robot capable of running, achieving an average speed of 10 km/h and a maximum speed of 12 km/h, while maintaining stable movement across various complex terrains [2][4] - The open-source community for "Tiangong" was launched in April 2025, featuring blogs, forums, technical documentation, open-source project showcases, and more, aimed at fostering global collaboration and accelerating technological breakthroughs in the embodied intelligence industry [1][6] Group 2: Collaboration and Development - Over a hundred partners have developed applications based on the "Tiangong" platform, leading to the emergence of capable robotic products like "Tiangong Walker" [4] - The RoboMIND dataset has been downloaded over 15,000 times from various platforms, including contributions from top universities and research institutions, indicating its significance in training embodied models [4][5] Group 3: Recognition and Awards - The RoboMIND dataset received the "Best Open Source Intelligent Breakthrough Award" at the OS2ATC 2025 conference and the "2025 Annual Dataset Construction Award" from leading industry organizations, highlighting its impact in the field [5] Group 4: Future Directions - The Beijing Humanoid Robot Innovation Center plans to continue open-sourcing a series of algorithms and tools related to embodied intelligence, including frameworks for reinforcement learning and various models, to further enhance the industry ecosystem [6] - The center aims to leverage open innovation to address common technological challenges and activate a global innovation network, contributing to the evolution of embodied intelligence towards generalization [6]

Core Viewpoint - Guangdong Province has introduced policies to support the development of open-source communities and ecosystems, with each receiving up to 8 million yuan in funding, aiming to establish itself as a global hub for artificial intelligence and robotics [1][2]. Group 1: Policy Measures - The policy emphasizes a shift from "single-item support" to "ecosystem construction," focusing on high-level platforms and ecosystem development to drive overall industry upgrades [1]. - Guangdong's AI core industry reached over 220 billion yuan last year, with more than 1,500 core enterprises, positioning it as a leader in the national AI and robotics sector [1]. Group 2: Ecosystem Development - The integration of open-source ecosystem construction into a collaborative system of "innovation chain, industry chain, capital chain, and talent chain" aims to transform Guangdong from a manufacturing province to a strong intelligent manufacturing province [2]. - The establishment of national-level data trading venues in Guangzhou and Shenzhen is intended to facilitate the open flow of data elements, providing high-quality data support for the open-source ecosystem [1]. Group 3: Challenges and Future Outlook - Challenges include avoiding the issue of open-source communities focusing more on construction than operation, and balancing data openness with security, privacy, and copyright protection [2]. - The cultivation of open-source communities and ecosystems is seen as both forward-looking and practically significant, with the potential to create a comprehensive innovation ecosystem in Guangdong that spans basic research, technology breakthroughs, commercialization, technology finance, and talent support [2].

Core Viewpoint - Alibaba's release of the Qwen3 model marks a significant advancement in AI technology, emphasizing efficiency and practicality over sheer parameter size, solidifying its position in the AI sector [1][2][5]. Technical Architecture - Qwen3 is the first "hybrid reasoning model" in China, integrating "fast thinking" and "slow thinking" to dynamically adjust computational resources based on task complexity [5][13]. - The model's pre-training dataset has expanded significantly, utilizing approximately 36 trillion tokens, nearly double that of its predecessor Qwen2.5, covering 119 languages and dialects [5][6]. Performance and Cost Efficiency - Qwen3 boasts a total parameter count of 235 billion, with activation requiring only 22 billion, leading to a substantial reduction in deployment costs [6][14]. - In various benchmark tests, Qwen3 achieved remarkable scores, including 81.5 in the AIME25 assessment and 95.6 in the ArenaHard evaluation, outperforming leading models like OpenAI-o1 and DeepSeek-R1 [6][10]. Open Source Strategy - Qwen3 is released under the Apache 2.0 license, allowing free commercial use, which positions it competitively against Meta's LLaMA series [7][9]. - The model family has surpassed 300 million downloads globally, establishing the largest open-source model ecosystem [7]. Commercialization and Market Impact - Alibaba's AI strategy focuses on practical applications rather than just laboratory metrics, aiming to create real-world value [11][17]. - The introduction of Qwen3 is expected to enhance the commercial viability of AI applications, particularly in sectors like finance, healthcare, and manufacturing [17]. Future Developments - Alibaba plans to invest 380 billion yuan in AI and cloud computing infrastructure over the next three years, indicating a commitment to deepening its AI capabilities [1][17]. - The model's flexible "thinking budget" feature allows developers to set maximum token consumption for deep thinking, catering to diverse performance and cost needs [13][14].

2025年是AI应用爆发的元年，当全球AI竞赛步入"中国时刻"，一场深刻的技术变革正悄然改写产业格局。在此关键节点，行业面临核心命题：如 何跨越AI技术到规模化应用的鸿沟？下一个颠覆性的AI超级应用将诞生于何处？ 4月18日，由36氪主办的2025 AI Partner大会于上海模速空间盛大启幕。本次大会以"Super APP来了"为主题，聚焦AI应用对千行百业的颠覆性变革。大会分 为"Super App来了"和"谁是下一个超级应用"两大篇章，覆盖"在AI世界中长大""2025卷AI就卷超级应用"等七大话题，涵盖10+场主题演讲、3场圆桌对话与 两大优秀AI案例企业名册发布环节，深度剖析AI技术如何重构商业逻辑、重塑产业格局，探索AI超级应用带来的无限可能。 当日，WeShop唯象总经理吴海波带来了《WeShop唯象的全球化之路：进化or淘汰》的主题分享。 以下为吴海波演讲内容，经36氪整理编辑： 感谢 36 氪的邀请！大家下午好，我是来自WeShop唯象的吴海波。今天，我想结合这两年WeShop的创业历程，分享一些实战中的观察与思考，希望能为身 处AI浪潮中的创业者带来启发。 WeShop唯象总经理吴海波 ...

人工智能 正以前所未有的速度，重塑着人类社会的每一个维度。 值得注意的是，在 A I 重塑世界 的同时，历史也来到大变局的关键时刻。在这个充满变化与不确定性的当下，焦虑与迷茫也常常伴 随左右，这种时候，不如静下心来读读书，给大脑"充充电"。 今年世界读书日，按照惯例，我们邀请高毅资产投研团队，订制2025年高毅"研选"书单， 我们以 "AI " 为锚点， 精选涵盖理论、实践与前沿趋势的书单 ，也延伸历史和世界，从过去看未来，探寻 人类发展之路。 当然， 仅靠纸质载体 无法 承载 当下 技术迭代的速度 ，我们还邀请投研团队分享他们的 "宝藏视 频博主" ——权威教授 以及业内专家的用心讲解 ，让抽象概念生动可触，让复杂技术通俗易懂。 当算力开始重写文明底层协议时， 当未来呼啸而来之际， 阅读 和学习， 本身就是最古老的反向驯 化。智者乐学， 祝 开卷有 " 毅 " 。 1. 《 黄仁勋：英伟达之 芯 》- 斯蒂芬·威特 推荐人 ：高毅资产资深基金经理 AI革命 01 推荐理由： 从显卡演化为GPU，软件革命与硬件革命的融合，在并行计算机上运行的神经网络，驱动着 人工智能革命。 2.《 泛函分析导论及应用》- 欧 ...

Cursor 用户：C/C++ 扩展已损坏！ 具体来看，不少 Cursor 用户在安装插件时频频报错。有开发者@ rektbuildr 在 GitHub 上面反馈称： 刚才编码时我收到此消息： C/C++ 扩展只能与 Microsoft Visual Studio、Visual Studio for Mac、Visual Studio Code、Azure DevOps、Team Foundation Server 以及后续的 Microsoft 产品和服务一起使用来开发和测试您的应用程序。 | Problems | Output | Debug Console | Terminal | Ports | C/C++ | > | × × = × × | | --- | --- | --- | --- | --- | --- | --- | --- | | Foundation Server, and successor Microsoft products and services to develop and test your applications. | | | | | | | | | The C/C+ ...

61岁的马云近半年来密集地在国内公开露面。 马云现身阿里 "高科技绝不仅是征服星辰大海，更应该呵护人间烟火。"4月10日，阿里巴巴创始人马云出现在阿里云 新财年启动会上并进行了演讲。他表示，未来不是让AI取代人类，而是应该让AI解放人类，更懂人 类，服务好人类，"我们不是去追求让机器像人，而是让机器去理解人类，像人类一样去思考，做人类 做不到的事情。" 点评：61岁的马云近半年来正密集地在国内公开露面，走入蚂蚁集团和阿里巴巴园区，关注夸克、闲 鱼、盒马、阿里云等业务的发展。AI是他最为关心的话题之一。马云强调AI必须帮助普通人，能否找 到技术变革中由人们生活便利需求催生的创新业务形态，对阿里云来说或许将是一项更长远、深刻的挑 战。 京东、盒马推"出口转内销"扶持计划 4月11日，京东宣布将推出2000亿元出口转内销扶持计划。未来一年内，京东将大规模采购不低于2000 亿元的出口转内销商品。 同一天，盒马宣布面向中国外贸企业开放入驻通道。盒马表示，将为外贸企业开通绿色通道，24 小时 专人加急处理，简化认证流程，缩短审核认证周期，盒马后续将上线"外贸专区"。对于有意愿创新研发 的外贸企业，盒马计划协同开发更贴合 ...

此次发布会上，华为、腾讯、深开鸿、美的库卡、广汽等9家企业发布了前沿成果。其中，深开鸿 发布基于开源鸿蒙的多机协同机器人统一操作系统——M-Robots OS1.0，可让多种机器人实现协同、兼 容。深开鸿CEO王成录称，机器人产业将从"单体智能"走向"群体智能"。未来，机器人产业和万物互联 的操作系统显然不可能是通过一两家企业就能做好，需要大家共建共享开源生态。 广汽第三代具身人形机器人GoMate在会上发布。广汽机器人团队负责人张爱民表示，GoMate计划 在2026年实现小批量生产，2027年逐步扩展至大规模量产。 乐聚机器人合伙人杨金介绍称，今年1月，该公司完成第100台全尺寸人形机器人的交付。目前人形 机器人在搬运等作业场景中还没达到与人一样的效率，但核心技术迭代速度很快，相信不远的将来人形 机器人有望实现与人一样的效率。 中新社广州4月9日电 (记者程景伟)记者9日从广东省人工智能与机器人产业创新产品与服务新闻发 布会上获悉，广东智能机器人产业主要布局于珠三角地区，并初步建立涵盖"大脑—五官—肢体"的智能 机器人全产业链。 据统计，2024年广东工业机器人产量超24万台(套)，同比增长31.2%，连 ...

撰 文 / 玖 零 设 计 / 琚 佳 2025年3月，理想汽车在中关村论坛上宣布开源自研的整车操作系统"星环OS"，直面传统 AUTOSAR架构的桎梏。这一动作被视为中国汽车工业在智能化时代的一次标志性事件。 这场技术革命的起点，源于前几年的一场"芯片荒"。 星环OS可兼容ARM、RISC-V、英飞凌TriCore等主流架构，支持地平线J6M、芯驰E3640等国产芯 片。这种"芯片自由"使得适配周期从6个月压缩至4周，适配成本降低83%，极大增强了供应链弹 性。这种灵活性使理想在2024年成功切换至国产RISC-V架构芯片，单车BOM成本下降900元，供应 链安全系数提升40%。 通过软硬件深度协同，星环OS在算力、存储和通信带宽资源使用上实现核心优化。典型案例是理 想L系列车型的控制器设计。通过将Orin X（SoC）与Tc397（MCU）两颗核心芯片的解耦管理，星 环OS实现了算力资源的动态分配：智驾任务可调用90%的Orin X算力，而在非驾驶场景下，这部分 算力又能被座舱娱乐系统复用。这种"一心多用"的虚拟化技术，使硬件利用率提升30%，每年节省 数十亿元BOM成本。 2020年，全球汽车芯片供应链 ...

技术的力量，开源的力量，初创生态的力量。 整理丨刘倩 ▲扫描上图中的二维码，可收听播客。《晚点聊 LateTalk》#100 期节目。欢迎在小宇宙、喜马拉雅、苹果 Podcast 等渠道关注、收听我们。 《晚点聊 LateTalk》是《晚点 LatePost》推出的播客节目。"最一手的商业、科技访谈，最真实的从业者思考。" 2025 年 1 月，农历春节也没有让模型竞赛丝毫减速。DeepSeek 发布开源推理模型 R1，以相对低的成本，在一些 Benchmark 上比 肩，甚至超越了 o1 的表现，在全球掀起了广泛讨论。 这期节目，我们邀请了 2015 年，在硅谷创立了 Fusion Fund 的投资人张璐，来和我们一起聊一聊，当前美国科技圈和硅谷语境中， 对 DeepSeek 等模型的讨论。 我们也延展聊了 DeepSeek-R1 和 o1 等推理模型打开的 Agent（智能体）应用空间；以及在美国的科技投资视野中，除了 AI，大家还 在关注什么。 Fusion Fund 曾投资 Grubmarket、Al 会议公司 Otter.ai 还有 Al 与医疗结合的公司 Subtle Medical 等。在 Al ...