智通财经APP获悉，智谱(02513)高开近15%，截至发稿，涨14.94%，报182.3港元，成交额1730.03万港 元，较招股价116.2港元已涨近60%。 消息面上，据智谱官方公众号消息，智谱与滴滴宣布达成战略合作，双方将围绕通用人工智能（AGI） 关键技术及其在出行领域的智能体应用开展前瞻性协同探索。双方将共同推进Agent场景落地和大模型 领域人才培养，深化出行场景的意图对齐与推理能力建设，推动Agent在更复杂业务场景中的验证与落 地。 根据公开资料，智谱是中国最早投身大模型研发的厂商，原创提出了基于自回归填空的通用预训练范式 GLM，率先发布了中国首个百亿模型、首个开源千亿模型、首个对话模型、首个多模态模型，以及全 球首个设备操控智能体(Agent)，形成了全面的模型体系，是国内罕有在原创技术路线上与全球顶尖水 平保持同步的厂商，因此也被誉为"中国OpenAI"。 ...

Core Viewpoint - The AI industry has entered a realization phase, with the competition among large model companies intensifying as they shift focus from scale to profitability [1] Company Overview - Zhiyu, established in 2019, specializes in foundational model development and has created a comprehensive model matrix covering language, code, multimodal, and intelligent agents, adapting to over 40 domestic chip types [3] - Zhiyu aims for AGI from its inception, distinguishing itself from competitors, and ranks first among independent general-purpose large model developers in China and second among all general-purpose large model developers globally based on projected 2024 revenue [4] Financial Performance - Zhiyu has begun to focus on profitability, launching a MaaS strategy in 2021, with nine of the top ten internet companies in China using its GLM model, making it the only startup with significant revenue from MaaS [5] - Revenue projections show significant growth, with expected revenues of 57.4 million yuan, 124.5 million yuan, and 312.4 million yuan from 2022 to 2024, reflecting a compound annual growth rate of 130%. In the first half of 2025, revenue reached 190 million yuan, a year-on-year increase of 325% [5] Losses and Margins - Despite rapid growth, Zhiyu's losses have also increased, with net losses of 143 million yuan, 788 million yuan, and 2.956 billion yuan from 2022 to 2024, and a net loss of 2.351 billion yuan in the first half of 2025 [7] - Gross margins have shown a declining trend, with rates of 54.6%, 64.6%, and 56.3% from 2022 to 2024, and a gross margin of 50% in the first half of 2025 [7] Investment and Market Outlook - Prior to its IPO, Zhiyu completed eight funding rounds, raising over 8.3 billion yuan from notable investors including Meituan, Ant Group, Alibaba, Tencent, Sequoia China, and Hillhouse [7] - According to CITIC Securities, Zhiyu's revenue has consistently doubled over the past two years, with expectations to exceed 1 billion USD in 2025. The domestic large language model market is projected to grow 20-fold over the next six years, with enterprise demand driving a trillion yuan opportunity [7] - The IPO of Zhiyu is seen as a valuation anchor for the industry, indicating a shift from explosive growth to stability, with capital focusing more on revenue than scale [8]

1. 证监会副主席陈华平1月11日在第三十届中国资本市场论坛上表示，将持续完善长钱长投的制度环境，合力推动各类中长期资金进一步提高入市规模比 例； 不断增强对科技创新企业服务的精准性、有效性，纵深推进科创板创业板改革，深化再融资改革；进一步提升监管执法有效性，持续健全投资者教 育服务和保护体系，坚持依法从严监管，突出打大、打恶、打重点，从严惩治各类恶性违法行为。 2.近日，清华大学基础模型北京市重点实验室发起AGI-Next前沿峰会，引发业界关注。 会议认为，大模型竞争已从"Chat"转向"Agent"阶段，重心从榜单 分数位移至真实环境的复杂任务执行。 行业预判2026年为商业价值落地元年，技术路径正向可验证强化学习（RLVR）演进。面对"中国反超"议题，领军 者持冷静态度，将领先概率评估为20%以内，认为中美在算力投入结构、新范式引领及toB生态上的本质差距。 3. 本周全球市场大事不断！中国2025年12月外贸数据、美国CPI及PPI数据将陆续公布；美国最高法院或公布关税裁决； 美联储将发布最新褐皮书，多位 美联储官员将密集发声；美股财报季开启，摩根大通等将披露最新业绩。重要会议方面，七国集团（G7）加 ...

本报讯(浦轩)近日，由江北新区企业中铁十四局大盾构公司研发的国内首个超大规模大盾构混合专家模 型——"深远大模型"，通过了中国信息通信研究院组织的"可信AI大语言模型"专项评估，并获4+级评 级，这是国内首个获国家级权威认证的盾构垂域大模型。 据悉，"深远大模型"是中铁十四局"数智盾构3.0"体系的核心，"为了让它更专业，我们给它输入超过80 万份与盾构施工相关的学术论文和领域专著，并首创性地构建了全球首个盾构施工领域的通识知识图谱 系统，把零散的专业知识系统化、结构化，为施工的智能化决策提供了核心知识支撑。"企业相关负责 人介绍，"深远大模型"是国内首个超大规模的大盾构领域混合专家模型，构建起"实时数据+专家经验 +行业知识+模型算法"相融合的技术体系，为盾构机的掘进提供"智慧大脑"。 例如，当一条过江隧道施工之前，可以将相关的地质环境数据告知大模型，它可以从数据库找理论，再 进行深度思考，最后给出客观、准确的前瞻性施工建议，并提示风险隐患，让施工的风险管控从"被动 应对"升级为"主动预防"。 此前，"深远大模型"及其应用体系已在杭州萧山机场线钱塘江隧道工程成功试点应用，并取得显著成 效。未来，盾构机掘进也 ...

第三方观察人士的关注点是前后脚上市的两家大模型公司的异同。文渊智库创始人王超向北京商报记者 表示，不管是智谱还是MiniMax，都抓住了AI上市第一波，尤其是中国资本市场先机的战略窗口。"实 际上两者的技术差距不是代际式的，有一定差距，但差距较小。普通投资人分不清两个公司各是什么背 景，但两家公司业务明显不同——一个注重C（用户）端，一个注重B（企业）端。" MiniMax的财报数据印证了这一点。2023年，MiniMax产生营收，当年AI原生产品给MiniMax贡献21.9% 的营收；2024年和2025年前9个月，这一部分营收占比分别是71.4%、71.1%。 大多数大模型公司在卖API（应用程序编程接口），MiniMax则把大模型应用推广到全球。1月9日，这 家强调多模态的大模型公司正式登陆港交所，成为继智谱之后第二家上市的国产大模型企业。与前一日 上市的智谱不同，MiniMax的营收主要来自全球消费者的钱包，产品包括虚拟AI扮演社交应用星野和 Talkie、AI视频平台海螺AI等。截至2025年9月，其产品覆盖超200个国家，累计用户达2.12亿，海外收 入占比超过70%。2025年前9个月，Mini ...

转自：北京日报客户端 1月10日，在由清华大学基础模型北京市重点实验室、智谱AI发起的AGI-Next前沿峰会上，清华大学教 授、院士张钹，加拿大皇家学院院士、香港科技大学荣休教授杨强，清华大学教授、智谱创始人唐杰， Qwen技术负责人林俊旸及腾讯"CEO总裁办公室"首席AI科学家姚顺雨等多位人工智能领域顶尖专家齐 聚，展开了关于AI未来与中国机会的脑力激荡。 "我对今年出现非常大的范式变革很有信心，在持续学习、模型记忆能力，甚至多模态领域，都有可能 出现新的范式变革。"智谱创始人唐杰的信心主要源自学术界的大模型研发正在跟上工业界的脚步。"两 年前，一些高校老师手上都没有卡（算力），如今很多高校都有了算力配置，也开始进行大模型架构、 持续学习的相关研究。"在唐杰看来，学术界已经加速铺开AI研究的土壤，有望孵化出新的创新种子。 与此同时，当投入产出比逼近瓶颈，追求"智能效率"的新范式必然诞生。 香港科技大学荣休教授杨强提出，学术界应着眼于研究那些工业界尚未解决的根本问题，如智能的上界 在哪里、如何平衡"记忆"与"推理"、应投入多少资源用于降低错误率和幻觉等问题。 圆桌讨论的其中一个议题直指国内外产学研各界关注的 ...

Group 1: IPO Highlights - Beijing Zhiyu Huazhang Technology Co., Ltd. (Zhiyu) officially listed on the Hong Kong Stock Exchange on January 8, 2026, under the stock code "2513," becoming the "first global large model stock" [2] - Zhiyu plans to issue 37,419,500 H-shares, with a public offering in Hong Kong receiving 1,159.46 times subscription and international offering receiving 15.28 times subscription, raising over 4.3 billion HKD at an issue price of 116.2 HKD per share [2] - Shanghai Tianshu Zhixin Semiconductor Co., Ltd. (Tianshu) also listed on January 8, 2026, under the stock code "9903," issuing 25,431,800 shares with a public offering subscription of 414.24 times and international offering of 10.68 times [5] - MiniMax Group Inc. (MiniMax) listed on January 9, 2026, under the stock code "0100," becoming the largest AI large model company by IPO scale in history [7] - Shenzhen Jingfeng Medical Technology Co., Ltd. (Jingfeng) listed on January 8, 2026, under the stock code "2675," issuing 27,722,200 H-shares with a public offering subscription of 1,091.94 times and international offering of 25.18 times [9] Group 2: Company Performance and Growth - Zhiyu has achieved a revenue growth from 0.57 million CNY in 2022 to 3.12 million CNY in 2024, with a compound annual growth rate (CAGR) of 130% [3] - Tianshu's revenue increased from 1.89 billion CNY in 2022 to 5.40 billion CNY in 2024, with a CAGR of 68.8% [6] - MiniMax's revenue grew from 3.5 million USD in 2023 to 30.5 million USD in 2024, representing a year-on-year increase of 782.2% [7] - Jingfeng's revenue for the first half of 2025 reached approximately 149 million CNY, a nearly 400% year-on-year increase [10] Group 3: Market Position and Future Outlook - Zhiyu is recognized as the largest independent large model vendor in China, with significant market advantages in the enterprise sector [4] - Tianshu's products have been deployed in over 900 instances across key sectors, indicating a strong market presence [6] - MiniMax has established a user base of over 2.12 million individuals and 130,000 enterprise clients across more than 200 countries [7] - Jingfeng's robotic surgical systems have been used in over 12,000 surgeries, indicating a growing integration into standard surgical practices [9]

Core Viewpoint - The current phase marks the early turning point of the AI application industry, with market sentiment reaching a beta stage, and the demand for AI applications is expected to rebound significantly in 2026 as foundational large models become more affordable and efficient [1][3]. Group 1: AI Application Demand - 2026 is anticipated to be the year of explosive demand for AI application agents, driven by continuous upgrades of global foundational large models throughout 2025, making them cheaper, smarter, and more reliable [3]. - The development logic of emerging industries follows a pattern: new supply products emerge, stimulating experimental demand, leading to qualitative changes in supply product performance, and eventually resulting in a consensus on demand that drives commercial value [4]. Group 2: Market Dynamics - The current internet era relies heavily on self-media for widespread exposure of new concepts, which accelerates the penetration of AI technology into the public consciousness and increases the frequency of mentions in institutional research reports [7]. - The AGI-Next summit highlighted the disparity in computational resources between the U.S. and China, with the former having superior hardware while the latter excels in algorithm optimization under resource constraints [8]. Group 3: Business Models and Applications - GEO (Generative Engine Optimization) is a new discipline emerging from the proliferation of generative AI, fundamentally differing from traditional SEO in its approach to information retrieval and optimization logic [9]. - The commercial value of GEO focuses on high-ticket scenarios such as legal and medical fields, contrasting with SEO's broader but lower-value applications [9]. Group 4: Industry Collaboration - Large model companies are unlikely to directly engage in GEO-related services to maintain the neutrality and reliability of their information, preferring to build ecosystems and provide technical interfaces for third-party service providers [11]. - The collaboration between large model companies and GEO service providers will ensure that advertising demands are met through a clear division of responsibilities, maintaining platform integrity while optimizing content [12]. Group 5: Market Sentiment and Future Outlook - The current market sentiment is at a turning point for AI applications, with a focus on emotional and funding-driven scenarios in the short term, transitioning to a phase of fundamental growth expectations later in the year [13]. - Key scenarios include AI marketing (GEO) and AI for science as primary emotional funding scenarios, while secondary scenarios like AI companionship and AI programming are expected to gain traction [13].

Group 1 - MiniMax, a domestic large model company, has exercised its overallotment option in full as of January 9 [1] - MiniMax officially listed on the Hong Kong Stock Exchange on January 9, with an issue price of 165 HKD per share [1] - Since its establishment in early 2022, MiniMax has developed a series of multimodal general large models, including MiniMax M2, Hailuo 2.3, Speech 2.6, and Music 2.0, which possess strong coding and agent capabilities, as well as extended context processing abilities [1]

Core Insights - The competition in large models has shifted from "Chat" to "Agent," focusing on executing complex tasks in real environments rather than just scoring on leaderboards. The industry anticipates 2026 as the year when commercial value will be realized, with a technological evolution towards verifiable reinforcement learning (RLVR) [2][4][5]. Group 1: Competition Landscape - The engineering challenges of the Chat era have largely been resolved, and future success will depend on the ability to complete complex, long-chain real tasks. The core value of AI is transitioning from "providing information" to "delivering productivity" [4]. - The bottleneck for Agents lies not in cognitive depth but in environmental feedback. Future training paradigms will shift from manual labeling to RLVR, enabling models to self-iterate in systems with clear right or wrong judgments [5][6]. - The industry consensus suggests that while China has a high chance of catching up in the old paradigm (engineering replication, local optimization, toC applications), its probability of leading in new paradigms (underlying architecture innovation, long-term memory) is likely below 20% due to significant differences in computational resource allocation [5][11]. Group 2: Strategic Opportunities - Opportunities for catching up lie in two variables: the global shift towards "intelligent efficiency" as scaling laws encounter diminishing returns, and the potential paradigm shift driven by academia around 2026 as computational conditions improve [5][19]. - The ultimate variable for success is not leaderboard scores but the tolerance for uncertainty. True advancement depends on the willingness to invest resources in uncertain but potentially transformative new paradigms rather than merely chasing scores in the old paradigm [5][10]. Group 3: Perspectives from Industry Leaders - Industry leaders express cautious optimism regarding China's potential to lead, with probabilities of success varying. For instance, Lin Junyang estimates a 20% chance of leading due to structural differences in computational resource allocation and usage [11][12]. - Tang Jie acknowledges the existing gap in enterprise AI lab research but bets on a paradigm shift occurring around 2026, driven by improved academic participation and the emergence of new algorithms and training paradigms [15][19]. - Yang Qiang believes that China may excel in toC applications first, drawing parallels to the internet history, while emphasizing the need for China to develop its own toB solutions to bridge existing gaps [20][24]. Group 4: Technological Innovations - The future of AI will require advancements in multi-modal capabilities, memory structures, and self-reflective abilities, which are essential for achieving higher levels of intelligence and functionality [68][70][73]. - The introduction of new optimization techniques, such as the MUON optimizer, aims to enhance token efficiency and long-context processing, which are critical for the performance of agent-based models [110][116]. - The development of linear attention mechanisms is expected to improve efficiency and performance in long-context tasks, addressing the limitations of traditional attention models [116]. Group 5: Future Directions - The industry is focused on distinguishing between scaling known paths through data and computational increases and exploring unknown paths to discover new paradigms [98][99]. - The potential for AI to participate in scientific research is anticipated to expand significantly, opening new possibilities for innovation and application [101].