Group 1 - The US and UK have preliminarily resumed negotiations on the "Tech Prosperity Agreement," valued at hundreds of billions of pounds, with initial progress in the civil nuclear energy sector, while AI and quantum computing collaborations remain stalled [1][2] - Discussions between senior officials from both countries have focused on civil nuclear technology cooperation and a joint nuclear fusion technology summit, indicating a thaw in a previously stalled cooperation framework due to bilateral trade tensions [1][2] - The partial resumption of the agreement is expected to benefit multinational energy companies by facilitating investments and project implementations, significantly reducing the regulatory approval timeline for nuclear energy projects from approximately three to four years to around two years [2][3] Group 2 - The "Tech Prosperity Agreement" was initially announced during President Trump's state visit to the UK in September last year, aiming to promote cooperation in AI, quantum computing, and nuclear energy, with a commitment of £31 billion from top US tech companies [2] - The UK government has engaged in constructive dialogues with US officials to revive the nuclear energy component of the agreement, with companies like Centrica and Rolls-Royce already making moves to enter the US market [2] - The agreement faced setbacks due to trade friction, particularly the US's dissatisfaction with the UK's willingness to address non-tariff barriers in broader trade negotiations, which previously led to a suspension of the agreement [3]

NEW YORK, Feb. 25, 2026 (GLOBE NEWSWIRE) -- Aeries Technology, Inc. (NASDAQ: AERT), a global leader in AI-enabled Global Capability Center (GCC) strategy and business transformation for private equity portfolio companies, today announced a new strategic advisory engagement with a leading global tax and financial advisory firm to support the expansion of its India-based GCC. Under the engagement, Aeries will deliver integrated advisory across people, processes, technology, and infrastructure as the client sc ...

Investment Rating - The report maintains a "Positive" investment rating for the industry [2] Core Insights - The industry is witnessing a threefold evolution in overseas large models, focusing on reasoning foundation, action implementation, and ecological reconstruction, with major players like Anthropic, OpenAI, Google, and OpenClaw leading the charge [7] - Domestic large model companies are breaking through through open-source foundations, efficiency optimization, and collaborative agent ecosystems, with firms like Z.ai, MiniMax, Kimi, Alibaba, and ByteDance showcasing diverse strategies [7] - The industry trend indicates a shift from generation to action, with competition centering on supply efficiency and ecological reconstruction, emphasizing task complexity and ROI as key growth drivers [7] Summary by Sections Overseas Large Model Evolution - Anthropic has completed a dual-version model iteration, focusing on enterprise workflows and enhancing coding and long-range agent capabilities [11] - OpenAI is concentrating on long-range task closure and tool execution, evolving its models to become collaborative productivity assistants [25] - Google has released Gemini 3.1 Pro to enhance reasoning capabilities and Lyria 3 to expand into audio creation, reinforcing its competitive edge [32][40] - OpenClaw is positioned as a self-hosted gateway, integrating multiple communication channels and supporting tool-based agents [43] Domestic Large Model Breakthroughs - Z.ai's GLM-5 aims to extend open-source model capabilities to complex systems and long-range agent tasks, with a focus on engineering deployment [48] - MiniMax's M2.5 emphasizes real-world productivity, optimizing costs and throughput to facilitate agent scalability [53] - Kimi's K2.5 leverages a multi-modal MoE architecture to enhance visual understanding and parallel agent execution [65] - Alibaba's Qwen3.5-Plus focuses on open-source upgrades and multi-modal transitions to drive agent scalability [73] - ByteDance's recent model releases aim to transition AI capabilities from dialogue to actual task execution [81] Industry Trends - The industry is transitioning from generation to action, with a focus on supply efficiency and ecological reconstruction, as evidenced by the increase in token processing volumes and the emergence of efficiency-friendly models [89]

Core Insights - The company, Envirotech Vehicles, Inc. (NASDAQ:EVTV), has accepted delivery of initial high-performance compute systems from AZIO AI Corporation [1] - The systems will be integrated into the company's modular artificial intelligence infrastructure platform [1] - The collaboration between the two companies is part of a strategic framework and ongoing due diligence process [1] Group 1 - The delivery marks a significant advancement in collaborative deployment efforts between Envirotech Vehicles and AZIO AI [1] - The installation, commissioning, and integration of the systems are scheduled to proceed as planned [1] - The partnership aims to enhance the company's AI capabilities through the new infrastructure [1]

Core Insights - Pengchuan Zhang, a prominent researcher from Tsinghua University, has joined OpenAI to focus on World Simulation and Robotics, indicating a strategic shift towards integrating visual perception and robotics technology [1][2][17] Group 1: Background of Pengchuan Zhang - Zhang graduated from Tsinghua University with a major in mathematics and later obtained a PhD in Applied and Computational Mathematics from Caltech in 2017, specializing in machine learning and deep learning applications in visual fields [3][4] - After completing his PhD, he worked at Microsoft Research as a principal researcher, leading projects in computer vision and multimodal intelligence [6][9] - Zhang has also held a part-time assistant professor position at the University of Washington since 2021, contributing to academic research alongside his industry roles [9] Group 2: Contributions at Meta - At Meta FAIR, Zhang led several groundbreaking projects, including the Segment Anything 3 (SAM 3) project, which provides a unified framework for object detection, segmentation, and tracking in images and videos [10][13] - He was also responsible for the Llama 3 and Llama 4 visual grounding projects, enhancing the models' capabilities in visual commonsense reasoning and complex scene understanding, significantly boosting Meta's generative AI competitiveness [13] Group 3: Industry Trends and Implications - Zhang's move to OpenAI is part of a broader trend where several high-profile researchers are transitioning to the company, driven by its advanced computational resources and foundational infrastructure for world modeling [16][17] - This shift suggests that OpenAI is making a significant investment in the "world model + physical intelligence" approach, which could lead to advancements in high-level robotic systems by 2026 [16][17]

21世纪经济报道记者 周潇枭 北京报道 2026年是"十五五"的开局之年，2月24日是春节后的第一个工作日，当天多个省份召开"新 春第一会"，以期更快地推动政策落地见效，推动经济稳中向好。 21世纪经济报道记者梳理发现，广东、江苏、山东、浙江、河南、湖北、湖南、安徽、江西、福建、辽宁、广西、海南等多个 省份召开"新春第一会"，对2026年的重点工作进行部署，包括构建现代化产业体系、千方百计扩大需求、加大投资力度、促进 民营经济发展壮大等。 多个省份都明确要强化人工智能技术突破和场景应用，还要加大新兴产业和未来产业投资布局，包括集成电路、自动驾驶、具 身智能、工业母机、低空经济、商业航天、脑机接口、生命科学等领域。另外，多个省份还明确将在2026年推出有力举措解决 拖欠企业账款问题。 加快布局智能时代 从地方"新春第一会"的布置来看，加快构建现代化产业体系是2026年的一项重点工作。 "随着人工智能广泛融入千行百业，在万物互联、平台赋能、数据驱动、软件定义等作用下，我们的生产方式、服务模式正在发 生革命性变化。大变革就是大机遇，无论是制造业还是服务业，希望广大从业者张开怀抱迎接智能时代的到来。"广东省委书记 黄 ...

Group 1 - The core issue involves the AI product "Yuanbao" generating offensive language in user-created New Year greeting posters, leading to user complaints [1][4] - Yuanbao's team acknowledged the problem, attributing it to abnormal outputs during multi-turn dialogues and has since corrected the issue while apologizing to users [4] - This incident is not isolated, as there have been previous reports of Yuanbao providing insulting responses when users requested code modifications, indicating a pattern of aggressive AI behavior [8] Group 2 - Yuanbao launched a significant marketing campaign during the Spring Festival, starting from February 1, with activities including cash red envelopes and other promotional events, aiming to enhance user engagement [8] - The marketing efforts resulted in substantial user growth, with daily active users exceeding 50 million and monthly active users reaching 114 million during the festival period [8] - Yuanbao's functionality updates were rapid, with 159 updates completed in 21 days, contributing to its rise to the top of the Apple App Store free chart [8]

In just eight months of existence, SolveAI has raised $50 million. That’s shorter than a pregnancy, but longer than an NFL season. The enterprise coding startup—incorporated in July by former Palantir engineer Steve Basher—has eleven employees, and raised its $5 million Accel-led pre-seed in August. Basher’s idea was simple: That an AI coding tool must be completely specific to the company using it.“AI’s somewhat useless without context,” Basher told Fortune. “Context is everything. So, our product is: what ...

Core Insights - The article highlights the unconventional path of Keller Jordan, who secured a position at OpenAI without a PhD or traditional research background, emphasizing the importance of open-source projects and practical contributions in the AI field [1][3]. Group 1: Keller Jordan's Journey - Keller Jordan graduated from UCSD in 2020 with dual degrees in mathematics and computer science, without having published any papers [5]. - His first job was at an AI content moderation startup, where he began to explore improvements in existing research [5]. - After reaching out to Google researcher Behnam for guidance, he collaborated on a project that led to a paper presented at ICLR [8]. Group 2: Contributions to AI Research - Keller's work on "NanoGPT speed run" significantly improved the training efficiency of Transformer models, achieving a 3.8 times increase in token efficiency, reducing the required tokens from approximately 10 billion to 2.7 billion [9][10]. - The design of the speed run was innovative, allowing for low-cost experimentation, with a single attempt costing as little as $8, making it accessible for individual researchers [12][13]. Group 3: Development of Muon - Keller developed an optimizer named Muon, which optimized the hidden layers of neural networks, achieving record training speeds for NanoGPT and CIFAR-10 [14][19]. - Muon demonstrated superior performance compared to the widely used AdamW optimizer, particularly as model sizes increase, indicating a potential breakthrough in AI model training [19]. Group 4: Entry into OpenAI - Keller officially joined OpenAI in December 2024, following the success of Muon in the developer community [20]. - He expressed a preference for continuing his research over publishing a paper, criticizing the prevalence of low-quality optimization papers in the field [21]. Group 5: Other Success Stories - The article also mentions other individuals who have successfully transitioned into major AI companies without traditional academic credentials, such as Sholto Douglas at Google DeepMind and Andy Jones at Anthropic, highlighting a trend of talent recognition based on practical contributions rather than formal publications [23][25][28].

掌管OpenAI命运的，竟是6个波兰天才？ OpenAI首席科学家Jakub Pachocki，接替Ilya之后，完成了后者多年来求而不得的突破。 奥特曼甚至专门发了一篇短文，提到OpenAI两位不可或缺的人物：Jakub Pachocki和Szymon Sidor。 奥特曼认为他们两人是天作之合，「OpenAI还没有他们两人没法解决的问题」，没有他们俩就没有现在的OpenAI。 你可能不知道的是两人在波兰读高中就已经认识，而网友Teortaxes抛出一张OpenAI团队合影，波兰人浓度爆表── OpenAI团队中六位顶级贡献者均为波兰人，从左到右分别为Jakub Pachocki、Łukasz Kaiser、Łukasz Kondraciuk、Szymon Sidor、Wojciech Zaremba、Jerry Tworek。 其中的Wojciech Zaremba，是OpenAI的十一位联合创始人之一。 在OpenAI成立之初，波兰工程师在团队中占据了相当大的比例，其中就有上面提到的Jakub Pachocki、Szymon Sidor、Łukasz Kondraciuk等人。 他们成为OpenAI最初 ...