Group 1: Generative AI Developments - OpenAI launched two new models, GPT-5.1 Pro and GPT-5.1-Codex-Max, with the former focusing on emotional and intellectual capabilities, while the latter is the first coding model to support a "compression" mechanism [1] - GPT-5.1-Codex-Max can autonomously work for over 24 hours, processing millions of tokens, with a 30% reduction in thinking tokens compared to previous versions, achieving a score of 77.9% on SWE-bench Verified [1] - Internal tests show that 95% of OpenAI engineers use Codex weekly, leading to a 70% increase in team Pull Request numbers [1] Group 2: Image Generation and Processing - Google introduced the Gemini 3 Pro Image preview, a reasoning model that performs internal reasoning before generating images [2] - This model supports 64K input tokens and 32K output tokens, capable of producing images with resolutions between 1K and 4K, and can combine up to 14 input images into one output [2] - It integrates Google search capabilities for up-to-date knowledge, excelling in complex multi-turn image generation editing and high factual accuracy creative tasks [2] Group 3: 3D Technology Advancements - Meta released the SAM 3D family, including SAM 3D Objects and SAM 3D Body, which can convert 2D image segmentation results into 3D models, even in the presence of occlusions [3] - SAM 3D features concept segmentation capabilities, achieving a 47.0% accuracy in the LVIS zero-shot segmentation task, surpassing the previous state-of-the-art (SOTA) of 38.5% [3] - SAM 3D Objects utilizes a 1.2 billion parameter flow-matching Transformer, outperforming other leading models by at least five times in direct comparison tests with human users [3] Group 4: Browser Innovations - QQ Browser's new version v19.8.5 introduces intelligent tab grouping and AI features for multitasking without interference [4] - The new web podcast feature supports AI podcasts and native reading with smart switching, allowing for precise 15-second navigation and five-speed adjustments [4] - The menu and functionality areas have been upgraded, with common tools like bookmarks and history easily accessible at the top, and all fixed function modules supporting drag-and-drop sorting [4] Group 5: Digital Identity Solutions - Second Me provides users with an independent ID and domain in the digital world, acting as an "AI ID" for expression and communication [5] - The product uses AI for precise matching of interests, focusing on finding individuals with detailed similarities rather than just shared interests, reducing communication costs in the industry [5] - Users can record fragmented notes and ideas, allowing their digital persona to continuously add memories and feedback for more natural and accurate expression [5] Group 6: Smart Wearable Technology - Lumia launched the world's first smart earrings, Lumia 2, weighing less than 1 gram and being five times smaller than AirPods, capable of real-time monitoring of head blood flow [7] - The product includes features for tracking sleep, body temperature, menstrual cycles, and overall health status, utilizing patented SwitchBack technology for compatibility with any earrings [7] - Lumia secured an additional $7 million in investment and $5.1 million in government funding, bringing total financing to $17.2 million, with its blood flow tracking technology published in top peer-reviewed journals [7] Group 7: AI Research and Development - Yann LeCun announced his departure from Meta after 12 years to pursue entrepreneurship focused on advanced machine intelligence (AMI) [8] - The new company's goal is to drive the next major AI revolution, enabling systems to understand the physical world, possess long-term memory, and exhibit reasoning capabilities [8] - Meta will partner with the new company, as LeCun emphasizes the importance of world model research, arguing that large language models (LLMs) cannot truly understand the physical world [8] Group 8: Space Computing Initiatives - NVIDIA has sent its H100 GPU into space for the first time, while Google plans to launch 81 satellites equipped with TPUs by 2027, intensifying the space computing competition [9] - China's CAS Tian-Suan has initiated the "Tian-Suan Plan," aiming to deploy a mega-scale space supercomputing center in sun-synchronous orbit, consisting of energy, computing, and communication modules [9] - By mid-2026, CAS Tian-Suan aims to achieve its first GPU supercomputing node in space, targeting a total computing power of 10 EOPS, powered by over 100MW of zero-carbon energy through flexible photovoltaic arrays [9] Group 9: AI Market Insights - NVIDIA reported a record Q3 revenue of $57 billion, with data center business revenue soaring 66% year-over-year to $51.2 billion, and provided a revenue guidance of $65 billion for the next quarter [10] - Jensen Huang refuted the "AI bubble" theory, highlighting a historic shift in computing paradigms from general CPUs to accelerated GPUs, with genuine and sustained demand for computing power [10] - The proportion of GPU-accelerated computing in the global TOP500 supercomputing list surged from 10% six years ago to 90%, with NVIDIA's gross margin around 70%, and global AI infrastructure investment projected to reach $3-4 trillion by 2030 [10]

大公司： "灵光"下载量破50万，冲上App Store总榜第七 36氪获悉，11月18日，蚂蚁集团正式发布全模态通用AI助手"灵光"，上线第一天下载量破20万，两天多 下载量破50万，第三天冲上App Store总榜第七，App Store免费工具榜第一。 空客启动第二阶段股票回购计划 空中客车公司11月20日宣布，正启动其于9月8日宣布的股份回购计划第二阶段，该计划旨在支持未来的 员工持股计划活动及股权激励计划。该计划将分多期在公开市场执行，截止日期为2026年1月16日，回 购股份上限为4140000股。计划首期于2025年10月31日完成，共回购2070000股。空客已委托一家投资公 司管理计划第二阶段的执行工作，该阶段将回购不超过2070000股股份，执行期自2025年11月20日起至 2026年1月16日止。 Meta首席科学家杨立昆官宣离职创业，瞄准高级机器智能 当地时间11月19日，担任Meta AI首席科学家的图灵奖得主杨立昆（Yann LeCun）在社交平台上宣布， 他即将离开工作了12年的Meta，开启事业的下一阶段：创业。这家新公司的研发核心将是"Advanced Machine In ...

Meta Platforms 证实 ，其首席人工智能科学家杨立昆（ Yann LeCun）将于年底离职创办自己的 AI 初创公司。 Meta 发言人表示，公司计划与杨立昆的初创企业建立合作关系。 杨立昆的新创公司预计将专注于所谓的世界模型，这是他长期深耕的 AI 研究领域。世界模型是一种试图通过图像、视频和其他多种数据来理解物理世界的 AI 形式，与主要在文本数据上训练的大语言模型不同。 在一篇面向 Meta 员工的内部帖子中， 杨立昆 表示他的新创公司将继续推进他在 Meta 旗下 AI 实验室 " 基础人工智能研究 " 和纽约大学（他目前仍担任教 授）所追求的先进机器智能工作。 " 这家初创公司的目标是带来 AI 领域的下一次重大革命：开发出能够理解物理世界、具备持久记忆、可以进行推理并规 划复杂行为序列的系统， " 杨立昆 在帖子中写道。 Meta 表示将与他的初创公司建立合作关系，并 " 能够获取其创新成果 " 。（ LeCun 随后在其 Facebook 页面发布了该声明。） Meta 当前的人工智能重点已转向 LLMs ，包括其 Llama 系列模型。 今年该公司投入数十亿美元高调招兵买马，组建了 ...

Core Viewpoint - Yann LeCun, a Turing Award winner, has announced his departure from Meta to start a new company focused on Advanced Machine Intelligence (AMI), aiming to revolutionize AI by enabling systems to understand the physical world, possess long-term memory, reason, and plan complex actions [1][8][14]. Group 1: Company Transition - LeCun's new venture will continue his research on "world models," which he believes are essential for AI to truly understand the physical world [8][27]. - Meta will act as a partner to LeCun's new company, supporting the AMI initiative, which has overlapping interests with Meta's business but also extends into other areas [8][28]. - The departure marks a significant shift in the AI landscape, as LeCun leaves a position he helped establish at Meta's FAIR (Facebook AI Research) amid internal cultural conflicts and strategic misalignments [17][27]. Group 2: Research Focus - The goal of the new company is to drive a major revolution in AI, focusing on systems that can understand the physical world and plan actions without extensive trial and error [8][24]. - LeCun has been a critic of large language models (LLMs), arguing that they lack true understanding of the physical world, and he aims to develop AI that can reason and plan using world models [19][27]. - Recent research contributions include the JEPA theory, which aims to create organized and actionable high-dimensional embedding spaces, seen as a potential pathway to achieving world models [25][27]. Group 3: Industry Impact - LeCun's transition to entrepreneurship at the age of 65 signifies a new exploration phase in AI, moving away from the constraints of corporate environments to pursue foundational scientific challenges [14][27]. - The departure of LeCun, alongside other key figures like Soumith Chintala, indicates the end of an era for Meta AI, highlighting the ongoing evolution within the AI research community [28].

Meta发言人确认杨立昆将离职。截至发稿，杨立昆尚未就此置评。(作者/箫雨) 根据这份与Meta同事分享的备忘录，杨立昆在周三宣布了这一消息。知情人士称，Meta计划与杨立昆 的新创业公司合作，但细节仍在敲定中。杨立昆在备忘录中表示，Meta"将成为这家新公司的合作伙 伴，并能够使用其创新成果"。 杨立昆正在创办一家新公司，聚焦他所称的"高级机器智能"(AMI)。这种AI形式通过视觉及其他感官信 息进行训练，能够让技术对物理世界进行预测。 "这家创业公司的目标是促成AI的下一次重大革命：开发出能够理解物理世界、具备持久记忆、拥有推 理能力并能规划复杂行动序列的系统。"杨立昆在备忘录中表示。他特别强调，该创业项目将是他在 Meta及纽约大学(他在该校担任教授)与同事们共同探索研究的延续。 杨立昆称："这项技术将在经济众多领域产生广泛应用，其中部分领域与Meta的商业利益存在交集，但 许多领域并不重合。通过独立实体来追求高级机器智能的目标，是将其广泛影响力最大化的一种途 径。" 凤凰网科技讯 北京时间11月20日，据彭博社报道，一份备忘录显示，"AI教父"之一、Meta首席AI科学 家杨立昆(Yann LeCun ...

Core Viewpoint - Meta has launched its new "world model" V-JEPA 2, aimed at enhancing AI's physical reasoning capabilities for better understanding and predicting the physical world [1][3][11] Group 1: V-JEPA 2 Overview - V-JEPA 2 is described as a "realistic abstract digital twin" that enables AI to predict the consequences of its actions and plan accordingly [1][3] - The model is 30 times faster than Nvidia's Cosmos model and has been open-sourced for developers to access and integrate into various applications [1][6][5] - V-JEPA 2 builds on the previous V-JEPA model released by Meta, further improving understanding and prediction capabilities [4] Group 2: AI Capabilities - The model provides AI with three core abilities: understanding, predicting, and planning, allowing it to create realistic internal simulations [3][17] - V-JEPA 2 can perform reasoning without the need for labeled video segments, distinguishing it from existing generative AI systems like ChatGPT [3][4] Group 3: Applications and Impact - The model is designed for real-time spatial understanding in AI-driven technologies such as autonomous vehicles, warehouse robots, and drone delivery systems [3][5] - Meta anticipates that V-JEPA 2 will pave the way for AI to operate autonomously in unfamiliar environments, potentially impacting sectors like healthcare, agriculture, and disaster response [18][19] Group 4: Competitive Landscape - The release of V-JEPA 2 is seen as a critical milestone in Meta's long-term AI roadmap, especially in the context of increasing competition with OpenAI, Microsoft, and Google [11][13] - The growing importance of world models in AI research is highlighted, with other companies like Google DeepMind also exploring similar projects [19] Group 5: Leadership and Strategy - Yann LeCun, Meta's Chief AI Scientist, emphasizes the need for AI to build models of how the world operates rather than merely mimicking human text [8][9] - Meta's CEO Mark Zuckerberg is reportedly taking a more hands-on approach to AI development, including significant investments in AI training data and the formation of new teams focused on achieving "superintelligence" [13][14][15]

责编 | 王启隆 出品丨AI 科技大本营（ID：rgznai100） 黄教主的演讲 感觉才没过几天，今年的 GTC 英伟达大会也即将迎来尾声了。 而今年比尔·达利则是对话"AI 教父" 杨立昆 （Yann LeCun），很有前后呼应的感觉。 但 GTC 并不只有黄仁勋和杨立昆，还有许多精彩的演讲与对话，比方说： ………… 接下来的一段时间， CSDN AI 科技大本营 将会在「 GTC 2025 大师谈 」栏目持续更新这些精华内容的全文整理，尽情期待。 比尔·达利 自己就在采访杨立昆之后进行了一场 演讲 ，系统性地讲解了英伟达 2024 一整年的四大项目进展，内容干货很多； OpenAI o1 作者 诺姆·布朗 （Noam Brown）和英伟达的 AI 科学家来了一场 对话 ，他认为现在 AI 圈最需要来一场革命的，就是这些五花八 门的 基准测试 （Benchmark），而且改这个东西还不需要花太多算力资源； 2018 年诺贝尔化学奖得主 弗朗西斯·阿诺德 （Frances Arnold）围绕 AI for Sciense 还有蛋白质工程进行了一场相当硬核的 圆桌对话 ； UC 伯克利教授 彼得·阿比尔 （P ...