今天分享的是：RGF薪酬观察2026：中国大陆篇（英文版） 报告共计：45页 RGF 2026中国大陆薪酬观察报告核心总结 《RGF 2026中国大陆薪酬观察报告》基于2025年1月至2026年1月期间50余万名候选人数据，聚焦五大行业薪酬趋势、就业市场动态及招聘核心挑战，为企 业薪酬策略制定与职场人士职业规划提供关键参考。报告中所指年度基本薪资均不含津贴、股票、奖金等额外福利。 2025年就业市场呈现"前三季度调整、第四季度复苏"的态势。招聘需求在前三季度温和收缩后，第四季度同比激增31%，彰显企业对2026年的战略信心；人 才供给全年保持稳健增长，第三季度增幅达32%，反映求职者信心与市场结构调整并行；面试接受率自年中起持续改善，第三、四季度分别同比上升14%和 7%，体现企业招聘意愿的强化。 薪酬趋势方面，2026年市场平均薪资增幅较2025年略有上升，折射经济与就业市场的稳定性。高需求岗位薪资涨幅显著，医疗行业销售、消费品行业新零售 相关岗位外部招聘薪资涨幅或达20%，自动驾驶领域算法专家更是有望达到30%。薪酬相关挑战中，"新聘员工薪资高于现有员工"的内部薪资倒挂问题最为 突出，占比33%，其次是薪 ...

Financial Data and Key Metrics Changes - In Q4 2025, sales reached $1,231.9 million, a 2.8% increase from the previous year, including a 1.4% organic growth and a $15.6 million unfavorable foreign currency impact [6][36] - Operating income (OI) for the quarter was $255.2 million, unchanged from last year, with an OI margin of 21.5%, down 60 basis points from the previous year [7][36] - Overall earnings for the corporation were $339.6 million, up 2.3% compared to 2024, with a total margin of 25.3% [7][39] - Quarterly EPS reached $4.94, an increase of 12 cents from $4.82 recorded last year [8][40] Business Line Data and Key Metrics Changes - The Commercial and Industrial (C&I) Group reported sales of $398.1 million, up $18.9 million or 5%, with an organic gain of $2.8 million and $7.9 million from favorable foreign currency translation [17][40] - The Tools Group had sales of $505 million, down from $506.6 million last year, but OI increased to $107.3 million from $106.9 million, with an OI margin of 21.2% [24][44] - The Repair, Diagnostics, and Information (RS&I) Group saw sales of $467.8 million, up $11.2 million compared to 2022, with an organic sales gain of $4.8 million [28][44] Market Data and Key Metrics Changes - The automotive repair market remains favorable, supported by an increasing average vehicle age of 12.8 years and rising household spending on vehicle repairs [8][9] - The C&I Group faced challenges from international headwinds, including the impact of the Ukrainian war and fluctuations in the Chinese economy [16][36] - Sales in critical industries rebounded despite previous government shutdown impacts, with military and defense applications showing recovery [41][76] Company Strategy and Development Direction - The company emphasizes continuous investment in product development, brand building, and enhancing its workforce to maintain competitive advantages [4][17] - Snap-on is focusing on leveraging its proprietary databases and software to improve repair efficiency and technician productivity [12][30] - The company is adapting to market demands by introducing innovative products, such as the NanoAxcess Cordless Lineup and ControlTech+ Torque Wrench, aimed at enhancing technician capabilities [20][22] Management's Comments on Operating Environment and Future Outlook - Management expressed optimism about the resilience of markets despite ongoing challenges, including tariffs and government shutdowns [4][53] - The company anticipates continued growth driven by investments in product innovation and a strong demand for vehicle repair services [34][53] - Management highlighted the importance of adapting to technician needs and market conditions to sustain growth [10][76] Other Important Information - The company has maintained a consistent dividend payment since 1939, with a 14% increase in dividends announced in November, marking the 16th consecutive year of increases [8][40] - The effective income tax rate for Q4 was 22.3%, slightly down from 22.5% in the previous year [39] Q&A Session Summary Question: Insights on the Tools Group's performance - Management noted that the Tools Group's performance was affected by external factors such as tariffs and government shutdowns, but gross margins improved by 150 basis points [61] Question: Performance of specific subsegments within the Tools Group - Management indicated that while tool storage was down, hand tools improved, diagnostics were down, and power tools saw growth [63] Question: Outlook for critical industries - Management expressed optimism for continued growth in critical industries, supported by an expanding custom kitting capacity and recovering military business [76]

Core Insights - Meta Platforms is optimistic about its new AI team and the upcoming launch of its core large model, Avocado, which has completed pre-training and is described as the company's most capable pre-trained foundational model to date [2][7] - The performance of Avocado has surpassed that of the best current open-source foundational models, and it matches top post-trained models in knowledge retention, visual perception, and multilingual capabilities, despite not yet completing the post-training phase [2][7] Group 1 - The internal memo indicates that Meta's AI model progress is optimistic but remains untested in the external environment, raising potential risks for the company [3][8] - Meta's previous AI model, Llama 4, underperformed, leading to a delay in its release and disappointment among developers regarding its actual performance [3][8] Group 2 - The setbacks in AI development prompted a significant restructuring of Meta's AI business, including the acquisition of Scale AI for $14.3 billion and the establishment of the Meta Superintelligence Labs led by Alexandr Wang [9] - Meta plans to increase its capital expenditure on AI, including computing costs, by approximately 73% in 2026, projecting a total of $115 billion to $135 billion [9] Group 3 - Avocado has demonstrated significant efficiency improvements, achieving a tenfold increase in computational efficiency compared to Maverick and over a hundredfold compared to Behemoth, which has not yet been released [4][9] - The efficiency gains are attributed to higher quality data acquisition, investment in model infrastructure, and the use of deterministic training methods, which are crucial for reducing energy consumption and costs in AI development [10] Group 4 - Recent public statements from Meta executives align with the positive tone of the internal memo, with CTO Andrew Bosworth highlighting similar efficiency improvements and CEO Mark Zuckerberg expressing confidence in the performance of upcoming models [5][10]

整理 | 华卫 氛围编码（Vibe coding）是否会摧毁开源生态系统？近日，多位知名研究人员在一篇预印本论文中 指出，从观测到的趋势及部分建模结果来看，情况可能确实如此。他们的警告主要集中在两方面：用 户互动逐渐从开源项目中剥离，同时启动一个新开源项目的难度大幅提升。 即便是热门开源项目，随着代码下载和文档查阅的需求被大语言模型聊天机器人的交互所替代，其官 网的访问量也出现下滑，项目商业规划推广、赞助募资和社区论坛运营的可能性也降低了。Stack Overflow 等社区论坛使用量的骤减也反映了这一点。 研究人员们最后的结论是：在氛围编码广泛应用的情况下，要维持开源软件目前的规模，就需要对维 护者的报酬方式进行重大改革。 而且，在氛围编码的相关补偿机制下，绝大多数开源项目都难以从中获益。 该论文指出，氛围编码降低了软件制作成本，但也改变了用户与软件生态系统的交互方式。在传统的 开源软件商业模式下，开发者会选择软件包、阅读文档，并与维护者及其他用户交流。而在氛围编码 模式下，AI 智能体可以端到端地选择、组合和修改软件包，人类开发者可能并不清楚使用了哪些上 游组件。 "AI 革命"or 人类智能的压力测试 ...

2026.02.05 Anthropic发布的面向企业法务的Claude Cowork 插件，可自动化完成合同审查、NDA 分类、合规追 踪、法律简报生成等核心法务工作，直接冲击法律信息服务与专业SaaS板块，二级市场情绪从"AI赋能 软件"迅速转向"AI取代软件"，担忧AI从工具直接升级为替代者，淘汰中间环节SaaS产品。 Palantir Technologies（PLTR.O）是近期该轮回调主要"受害者"，即使官方在本周早些时候发布了一份高 于华尔街预期的财报，股价却最终收跌11.62%。 面对该轮踩踏，多位企业高层对外否认担忧。谷歌母公司Alphabet财报电话会上，CEO桑达尔·皮查伊并 不认可"AI工具可能会损害大型软件SaaS公司定价权"这一观点。他表示，当前最关键的问题是计算能力 相关的限制——包括电力、土地、供应链等方面的约束，以及如何扩大产能以满足当前激增的需求、确 保长期投资的合理性，同时在这一过程中提升效率、保持行业领先水平。 英伟达创始人CEO黄仁勋近日在与思科CEO对谈中表示，AI不会做工具，而是会用工具。就像人不会 重新发明螺丝刀，AI 也会直接使用已经成熟的软件，通过调用接口、 ...

2026.02. 05 本文字数：1562，阅读时长大约3分钟 Anthropic发布的面向企业法务的Claude Cowork 插件，可自动化完成合同审查、NDA 分类、合规追踪、法律简报生成等核心法务工作，直接冲击法律 信息服务与专业SaaS板块，二级市场情绪从"AI赋能软件"迅速转向"AI取代软件"，担忧AI从工具直接升级为替代者，淘汰中间环节SaaS产品。 Palantir Technologies（PLTR.O）是近期该轮回调主要"受害者"，即使官方在本周早些时候发布了一份高于华尔街预期的财报，股价却最终收跌 11.62%。 面对该轮踩踏，多位企业高层对外否认担忧。谷歌母公司Alphabet财报电话会上，CEO桑达尔·皮查伊并不认可"AI工具可能会损害大型软件SaaS公司定 价权"这一观点。他表示，当前最关键的问题是计算能力相关的限制——包括电力、土地、供应链等方面的约束，以及如何扩大产能以满足当前激增的需 求、确保长期投资的合理性，同时在这一过程中提升效率、保持行业领先水平。 英伟达创始人CEO黄仁勋近日在与思科CEO对谈中表示，AI不会做工具，而是会用工具。就像人不会重新发明螺丝刀，AI 也会直 ...

2月4日，Nature刊登了一项由华盛顿大学与艾伦人工智能研究所主导研发的科研成果——OpenScholar。这是全球首个专为科学研究设计的、全开源的检 索增强生成（RAG）语言模型。它不仅能精准检索、拒绝幻觉，更能生成高质量的引用式回答。 OpenScholar 的引文准确率与人类专家相当，虽然仍需进一步优化，但该工具有望帮助科学家处理复杂且日益繁重的科学文献综述任务。 论文链接：https://www.nature.com/articles/s41586-025-10072-4 尽管大语言模型（LLM）在许多领域表现出色，但在科研辅助任务中仍面临严峻挑战：随着科学文献总量的快速增长，模型难以跟上最新进展，且常伴 随严重的"幻觉"现象。实验数据显示，GPT-4o 在引用科学文献时，产生错误引用的比例高达 78% 至 90%。 OpenScholar 通过整合 4500 万篇开放获取论文和独特的自反馈机制，实现了精准的文献检索与准确的引用生成，有效解决了现有模型在科学知识合成中的 准确性与可信度问题。 首个全开源的科学文献综述AI系统 OpenScholar 是一个专门为科学研究任务设计的检索增强语言模型。它 ...

根据弗若斯特沙利文的数据，按2024年营业收入计，本公司在中国产业级AI智能体提供商中位列第 五，市场份额为2.8%;并在中国以图为核心的AI智能体提供商中位列第一，市场份额约为50%。根据弗 若斯特沙利文的数据，公司是行业内中国首家通过知识图谱有效减少大模型幻觉的企业，此现象指人工 智能模型产出看似连贯合理却实为错误、荒谬或事实不准确的输出内容。一家能够在事实准确度、幻觉 率及复杂逻辑推理能力叁大维度构建核心竞争力与技术壁垒的企业，能成功缓解大语言模型的幻觉问 题;在事实准确度维度，其可确保输出事实具备高可靠性;在幻觉率维度，其输出内容中无法验证或经验 证属错误的占比较低;在复杂逻辑推理能力维度，模型在深度推演过程中产生逻辑断裂或捏造推理链等 幻觉现象的概率更低。凭借公司在多应用场景知识图谱构建领域的专长，及自多应用场景中积累的经 验，再加上由公司客户提供的受控且经过验证的数据集，本公司在减少AI系统中的幻觉方面实现突 破。此项技术突破使公司能够开发智能体，并为企业提供人工智能解决方案。通过融合大语言模型强大 的推理规划能力、知识图谱的精确性与可解释性，加之公司的专业知识及落地实施能力，本公司开发的 智能体 ...

Core Viewpoint - The company, Haizhi Technology Group, is set to launch an IPO from February 5 to February 10, 2026, offering 28.03 million H-shares globally, with a maximum price of HKD 28.0 per share, aiming to raise approximately HKD 648 million for various strategic initiatives [1][6]. Group 1: Company Overview - The company is a leading enterprise in China focused on developing Atlas graph solutions and industrial-grade AI solutions through graph-model fusion technology [2]. - The Atlas graph solution is one of the core products, enabling efficient integration, processing, and management of data from multiple sources [2]. Group 2: Market Position and Performance - According to Frost & Sullivan, the company ranks fifth among industrial-grade AI providers in China with a market share of 2.8%, and first among graph-centric AI providers with a market share of approximately 50% [3]. - The company's revenue grew from RMB 313 million in 2022 to RMB 503 million in 2024, representing a compound annual growth rate (CAGR) of 26.8% [3]. Group 3: Technological Advancements - The company has developed a breakthrough technology that effectively reduces the "hallucination" phenomenon in large language models, ensuring high reliability in output facts and low rates of unverifiable errors [3]. - The Atlas intelligent agent, launched in September 2023, enhances the reasoning capabilities of large language models in specific industry contexts [4]. Group 4: Strategic Focus and Investment - The company is focusing on parallel development of the Atlas graph solution and Atlas intelligent agent, which allows it to capture growth opportunities during different stages of digital transformation [5]. - The company has entered cornerstone investment agreements with several investors, agreeing to raise approximately USD 15 million under certain conditions [6]. Group 5: Use of Proceeds - The net proceeds from the global offering are planned to be allocated as follows: approximately 45% for R&D to enhance graph-model fusion technology, 20% for optimizing the Atlas intelligent agent, 15% for deepening customer collaboration, 10% for strategic investments, and 10% for working capital and general corporate purposes [6].

格隆汇2月5日丨海致科技集团(02706.HK)发布公告，公司拟全球发售2803.02万股H股，中国香港发售股 份280.32万股，国际发售股份2522.7万股(以上可予重新分配)；2026年2月5日至2月10日招股，预期定价 日为2月11日；发售价将为每股发售股份25.6-28.0港元，H股的每手买卖单位将为200股，招银国际、中 银国际及申万宏源香港为联席保荐人；预期H股将于2026年2月13日开始于联交所买卖。 公司是中国领先的企业，专注于通过图模融合技术开发Atlas图谱解决方案及产业级智能体并提供产业 级人工智能解决方案。知识图谱是一种图谱结构化数据模型，将资料呈现为相互连结实体、其关系与属 性的网络；而大语言模型则是通过自监督机器学习，在海量文本资料上训练而成的AI模型，用于执行 自然语言处理任务，例如语言生成。Atlas图谱解决方案是公司于往绩记录期间的核心产品之一。Atlas 图谱解决方案基于公司行业领先的图计算与数据分析能力构建，包含DMC数据智能平台、Atlas知识图 谱平台及AtlasGraph图数据库。该等解决方案使企业能够高效整合、处理及管理来自多个来源的数据， 构建及分析知识图谱， ...