公开信息显示，阶跃星辰于2025年7月联合近10 家芯片及基础设施厂商发起"模芯生态创新联盟"，旨在 打通芯片、模型与平台之间的技术壁垒，通过联合优化提升算力利用效率，加速大模型在各行业场景中 的应用落地。业内认为，随着推理模型成为主流，模型与算力的深度协同将成为推动大模型规模化应用 的重要路径。 责任编辑：宋雅芳 新浪科技讯 2月2日上午消息，阶跃星辰发布新一代开源 Agent 基座模型 Step 3.5 Flash。据介绍，该模 型面向实时 Agent 工作流场景打造，兼顾推理速度、智能水平与使用成本，在单请求代码类任务上， Step 3.5 Flash 最高推理速度可达每秒350个 token。 新浪科技讯 2月2日上午消息，阶跃星辰发布新一代开源 Agent 基座模型 Step 3.5 Flash。据介绍，该模 型面向实时 Agent 工作流场景打造，兼顾推理速度、智能水平与使用成本，在单请求代码类任务上， Step 3.5 Flash 最高推理速度可达每秒350个 token。 此外，Step 3.5 Flash 采用稀疏MoE架构，每个 token 仅激活约110 亿个参数（总计1960 亿参数） ...

港股大模型概念股MINIMAX-WP涨幅扩大至20%，创上市来新高，报590港元，上市来累计涨超 250%。 ...

"我主要是来滑雪的，抽空聊聊AI。"1月23-25日的2026崇礼论坛上，360集团创始人周鸿祎发布演讲， 先幽默开场，紧接着抛了一个关键论断——2026年，预计全世界至少出现100亿个智能体。 他认为，当前AI正经历"双线进化"，大模型迭代升级的同时，智能体也在快速发展。作为AI能力落地的 关键载体，智能体之所以关键，在于其能够深入理解垂直领域的专业知识，并与人类智力和劳动力高效 协同，从而推动生产力跃迁。 阿里国际数字商业集团副总裁、阿里巴巴设计委员会理事长杨光指出，AI时代，设计师不缺先进生产 力工具。他认为，伴随AI技术能力的涌现，设计师的能力体系也需要变革，技法重要性下降，审美和 工具应用愈发关键。 杨光以自身的跨界创作为例强调，不仅需要与AI协作，更要学会训练出属于自己的AI，让每个人都有 机会成为突破边界的创作者。 面壁智能联合创始人兼CEO李大海表示，模型能力与商业落地必须"两条腿走路"，面壁智能早早锚定端 侧模型与垂直场景，避免与大厂在通用搜索红海正面竞争，加速商业化闭环。 他认为，端侧模型是物理世界AGI的必由之路。 站在2026年的起点展望未来，大模型的竞争，进入拼智能、拼落地效能的深水 ...

Core Insights - The AGI-Next summit gathered prominent figures in the AI industry to discuss new paradigms, challenges, and opportunities for Chinese large model companies [1] - Yao Shunyu, Tencent's Chief AI Scientist, highlighted the distinct characteristics of the To C and To B markets in the large model sector, emphasizing the need for vertical integration in consumer applications and the premium placed on high-performance models in enterprise settings [4][5] Group 1: Market Dynamics - The To C market does not require high intelligence most of the time, with applications like ChatGPT serving as enhanced search engines [4] - In contrast, the To B market shows a strong willingness to pay for top-tier models, with companies willing to pay $200/month for premium models while showing little interest in lower-tier options [4] - Yao noted that the gap between strong and weak models in the To B market is widening, as errors from weaker models incur significant hidden costs [4] Group 2: Future AI Paradigms - Yao emphasized that future competitiveness will hinge on capturing context rather than merely increasing model parameters, suggesting that understanding user context is crucial for providing relevant responses [5] - He also pointed out that the development of autonomous learning signals is already underway, although current models lack the pre-training capabilities of leading companies like OpenAI [6] - The potential for new paradigms in AI is linked to the convergence of academic and industrial innovations, with universities increasingly equipped with computational resources [9] Group 3: AI Agent Development - The evolution of AI Agents is seen as a key change in the AI industry, with a framework proposed that outlines the transition from human-defined goals to AI autonomously defining its objectives [11] - The challenge of addressing long-tail demands is highlighted as a significant value proposition for AGI [11] - Commercialization of AI Agents faces challenges related to value, cost, and speed, with a need for Agents to solve meaningful human tasks without incurring excessive costs [12]

Core Insights - The 19th Huaxia Institutional Investor Annual Conference and Huaxia Financial (Insurance) Technology Forum was held in Beijing, focusing on the theme of "Vitality and Resilience, Innovation and Empowerment" [2] - The conference aimed to address contemporary challenges, build consensus for development, and explore future pathways [2] Group 1: Technology and Innovation in Securities - Wang Mian from CICC highlighted the importance of enhancing two core capabilities in the securities industry: value assessment for price discovery and effective risk management [2][4] - CICC's one-stop investment research service platform, "CICC Insight," aims to systematically convert years of research and frameworks from over 300 analysts into structured information to assist investment decision-making [3] Group 2: Risk Management in Global Capital Markets - CICC reported that over 30% of its revenue comes from international business, emphasizing the need for robust risk management capabilities when Chinese financial firms operate abroad [4] - The company has developed an integrated risk and capital measurement system for derivatives, significantly improving computational efficiency, reducing the time for risk capital measurement from 4-6 hours to under 1 hour [4]

Core Insights - Large models in AI are systematically reshaping the financial industry, creating a new ecosystem of efficient, intelligent, inclusive, and personalized financial services [1] - The company aims to leverage AI to enhance operational efficiency, reduce information asymmetry, and empower underserved customer segments [1][2] - The "5+n strategy" will guide the company's AI initiatives, focusing on five core application scenarios by 2025 [3] Group 1: AI Strategy and Implementation - The company plans to deepen human-machine collaboration by equipping employees and clients with dedicated AI assistants, expanding AI applications in core business areas [2] - The "5+n strategy" emphasizes a goal-driven approach, integrating AI into various business functions, including wealth management and research services [3] - The company has achieved significant milestones in AI infrastructure, data governance, and application exploration, with a structured work plan for gradual implementation [3] Group 2: Client Value and Service Enhancement - The company focuses on enhancing client value through AI-driven research services and personalized wealth management, exemplified by the launch of the "CICC Insight Model" [4] - The "CICC Insight Model" serves over 1,600 institutions and 44,000 clients, providing 24/7 digital research assistance with advanced AI capabilities [4][5] - The "CICC Smart Reading Model" has improved document quality in investment banking, achieving a 90% detection rate for semantic errors [5] Group 3: Internal Efficiency and Risk Management - AI tools like the "CICC Smart Assistant" and "CICC Smart Reading" optimize internal processes, enhancing compliance and operational efficiency [6] - The company has automated the review of thousands of internal documents, significantly reducing the workload on compliance personnel [6] Group 4: AI Ecosystem and Infrastructure - The company adopts a strategy of self-control in AI development, ensuring complete mastery over core technologies while collaborating externally [7] - Key platforms like "Tiansuan" and "Jiuzhang Zhiqi" have been established to support AI training and application development, with over 1,300 AI applications launched [7] Group 5: Challenges and Future Outlook - The company identifies significant challenges in AI adoption, including data quality, model interpretability, and system compatibility [9][10][11] - The company anticipates that AI will fundamentally alter the business model and competitive landscape of the securities industry over the next 3-5 years [11][12] - Future plans include integrating AI with big data technologies to transition from digital to intelligent transformation, focusing on data, applications, and computing power [12]

Group 1 - The article announces the recruitment of 10 partners for the autonomous driving sector, focusing on course development, paper guidance, and hardware research [2][5] - The recruitment targets individuals with expertise in various advanced models and technologies related to autonomous driving, such as large models, multimodal models, and 3D target detection [3] - Candidates are preferred to have a master's degree or higher from universities ranked within the QS200, with priority given to those with significant conference contributions [4] Group 2 - The benefits for partners include resource sharing related to job seeking, doctoral studies, and overseas study recommendations, along with substantial cash incentives [5] - Opportunities for collaboration on entrepreneurial projects are also highlighted [5] - Interested parties are encouraged to contact via WeChat for further inquiries regarding collaboration in the autonomous driving field [6]

Group 1 - The article announces the recruitment of 10 partners for the autonomous driving sector, focusing on course development, paper guidance, and hardware research [2][5] - The recruitment targets individuals with expertise in various advanced technologies such as large models, multimodal models, and 3D target detection [3][4] - The article highlights the benefits of joining, including resource sharing for job seeking, PhD recommendations, and substantial cash incentives [5][6]

Group 1 - The core viewpoint of the forum is that the integration of data and AI serves as a dual engine driving innovation and growth in the intelligent era [1][2] - Current challenges in large model development include a "data wall" dilemma, where the contribution of unlabeled data to model performance is diminishing, leading to a need for a shift from expert-driven data science to quantitative and self-evolving stages [1] - A practical example shared indicates that filtering high-quality data from a vast dataset can significantly enhance model accuracy, with a reported 1.7% improvement in domain question-answering tasks by using only 20% of high-quality data from 10 billion tokens [1] Group 2 - Emphasis on data quality is crucial, with a focus on both human and machine experiences to enhance large model performance [2] - The global AI data industry is undergoing a significant transformation from labor-intensive to technology and knowledge-intensive models, showcasing how high-quality data can benefit various industries through real-world applications [2] - High-quality datasets should meet the VALID² criteria (vitality, authenticity, large sample size, completeness, diversity, high knowledge density), indicating a systematic reconstruction of methodologies, infrastructure, and industry ecology [2]

Group 1: Industry Overview - The rapid development of large models is pushing artificial intelligence (AI) technology to a new stage, transitioning from weak AI focused on specific tasks to strong AI capable of handling complex tasks [1] - According to IDC, the market size for large model development platforms in China is expected to reach 1.69 billion RMB in 2024 [1] - The AI computing power market in China is projected to be approximately 19 billion USD in 2024, growing to 25.9 billion USD in 2025, representing a year-on-year growth of 36.32%, and reaching 55.2 billion USD by 2028 [1] Group 2: Company Profile - Cambrian - Cambrian is a globally recognized emerging company in the smart chip field, focusing on the research and development of AI chip products and technological innovation [2] - The company has developed several leading smart processors and chip products, including the Cambrian 1A, 1H, and 1M series for terminal scenarios, as well as cloud-based intelligent acceleration cards based on various chips [2] - Cambrian's smart chips and processors efficiently support large model training and inference, visual processing, speech processing, natural language processing, and recommendation systems, compatible with mainstream open-source large models [2] Group 3: Smart Computing Cluster System - Cambrian's smart computing cluster system integrates self-developed boards or intelligent complete products with partner-provided server, network, and storage devices, along with the company's cluster management software [3] - The smart computing cluster focuses on the application of AI technology in data centers, providing comprehensive hardware and software solutions for clients with relatively weak technical capabilities in AI application deployment [3]