Core Insights - The competition for the title of "the world's first stock of large models" is unfolding between two Chinese companies, Zhipu AI and Minimax, with their prospectuses revealing critical business details beyond the title itself [1] - The prospectuses highlight three key dimensions: the quality of real revenue, the pressure of computing costs on profits, and the survival period supported by cash reserves [1] - The prospectuses also show that the two companies are following distinctly different paths in their business models, representing potential breakout directions for Chinese large model companies in the current market environment [1] Financial Performance - Zhipu AI reported a revenue of 312 million CNY (approximately 31.2 million USD) for 2024, while Minimax reported about 30.38 million USD (approximately 216 million CNY) [4] - Zhipu AI's gross margin stands at 56.3%, significantly higher than Minimax's 12.2%, indicating a stark difference in their business models [3][5] - Both companies are in a "high investment for growth" phase, with Zhipu AI experiencing a net loss of 2.96 billion CNY and Minimax a loss of 465 million USD [4] User Data and Market Focus - Zhipu AI primarily targets B-end enterprises and developers, boasting over 12,000 enterprise clients and 4.5 million developers, while Minimax focuses on C-end users with 27.6 million monthly active users [5] - Zhipu AI's revenue structure shows a heavy reliance on localized deployments, contributing 85% of its revenue, while Minimax's revenue is increasingly derived from AI-native applications, which account for about 71% [14][15] Product Structure and Innovation - Zhipu AI's core product is the GLM series, which competes with OpenAI, while Minimax's abab series focuses on interactive and multi-modal capabilities [5] - Zhipu AI's revenue model is primarily based on privatized deployment (84.5%), whereas Minimax operates on a subscription model for AI applications [5] Globalization and Market Strategy - Minimax has shifted its revenue sources from 80% domestic to 73% international, indicating a strong commitment to globalization [14] - Both companies are exploring international markets, with Zhipu AI generating revenue in Southeast Asia and Minimax aggressively pursuing global expansion [17] Cash Reserves and Financial Viability - As of December 31, 2024, Zhipu AI holds approximately 2.457 billion CNY in cash, while Minimax has about 1.046 billion USD, indicating a significant difference in their financial stability [23][24] - Zhipu AI's cash reserves can sustain operations for about 11 months without new financing, while Minimax's reserves can last approximately 37.5 months [24] Risk Factors and Challenges - Both companies express concerns about high survival thresholds in the rapidly evolving AI technology landscape, with Zhipu AI focusing on supply chain security and geopolitical issues, while Minimax highlights its status as an uncommercialized company facing high bankruptcy risks [25][26] - The reliance on expensive computing resources and the significant R&D expenditures exceeding revenues pose financial challenges for both companies [20][21] Valuation and Market Position - The valuation of Zhipu AI and Minimax is approximately 40 billion CNY and 4 billion USD, respectively, which is significantly lower than that of OpenAI, indicating a potential undervaluation in the market [27][28] - The competitive landscape is intensifying, with established players like Alibaba and ByteDance posing significant challenges to the market position of Zhipu AI and Minimax [29]

来源：智通财经 "港股最大的优势就是便宜。"近日，国海富兰克林基金权益投资总监赵晓东在接受智通财经采访时表 示，港股目前估值处于低位，且大部分大企业注重股东回报，在分红、回购方面更为积极，公司治理也 更为规范，"当下最显眼的机会在港股。" 从资金面看，赵晓东认为，南下资金持续流入成为港股重要支撑，尤其在境内低利率环境下，港股高分 红资产吸引力凸显。若美国降息周期开启或人民币进入升值通道，港股有望迎来进一步的估值修复动 力。 展望未来，赵晓东认为两大外部变量可能进一步提振港股：一是若美国开启降息周期，有望带动全球资 金回流港股市场；二是若人民币进入升值通道，则以人民币计价的资产吸引力将上升，尤其有利于现金 流稳定的高分红标的。 当前为地产行业逆向布局的窗口期 针对市场普遍谨慎的地产行业，赵晓东认为其虽仍面临基本面压力——房价下行、供给过剩及政策对二 手市场的冲击尚未完全消散，但行业已处于底部区域，反而提供了逆向布局的窗口。他预计，明年一季 度或将迎来稳定地产的政策，其目标在于稳住市场信心，而非助推房价上涨。 在具体配置方向上，赵晓东当前重点关注港股中的地产、银行及互联网板块。 他指出，地产行业目前已处于底部区域 ...

Core Viewpoint - MiniMax is set to become the shortest time-to-IPO AI company if it successfully lists on the Hong Kong Stock Exchange, marking a significant milestone for Chinese AI startups in a challenging funding environment [3][10]. Company Summary - MiniMax plans to issue approximately 33.58 million shares for its overseas listing [2]. - The company has shown a narrowing net loss and a positive gross margin trend, indicating a transition to a healthier growth path [3][9]. - As of September 30, 2025, MiniMax's cash reserves total $1.102 billion, with cumulative financing exceeding $1.5 billion, demonstrating efficient capital utilization [7][9]. - The adjusted net losses for MiniMax from 2022 to 2025 show a decreasing trend, with a significant reduction in loss per unit of revenue [8][9]. - MiniMax's revenue growth is notable, with a year-on-year increase of 174.7% in the first nine months of 2025, while adjusted net losses only slightly increased by 8% [9][26]. - The company has a strong focus on both B2C and B2B markets, with B2C revenue accounting for over 71% and B2B gross margin reaching 69.4% [16]. Industry Summary - The successful IPO of MiniMax could serve as a model for other AI startups, demonstrating a sustainable path to public markets without heavy reliance on capital infusion [10][19]. - The AI industry is experiencing intense competition for funding, with established giants and startups vying for resources, making MiniMax's approach particularly relevant [4][5]. - MiniMax's strategy emphasizes a balanced growth model, focusing on organizational efficiency and human capital rather than solely on high capital expenditure [21][25]. - The company has made significant strides in international markets, with over 70% of its revenue coming from overseas, showcasing its global reach [18][19]. - MiniMax's unique organizational structure, with a high percentage of R&D personnel and a flat management hierarchy, enhances decision-making speed and resource allocation efficiency [21][23].

2025年12月22日，中国证监会国际合作司正式发布北京智谱华章科技股份有限公司境外发行上市及境内未上市股份"全流通"备案通知书，这一关键监管许 可的落地，标志着智谱华章的港股IPO进程迈入实质性阶段，公司向"全球大模型第一股"的目标发起最后冲刺。 | 中国证券监督管理委员会 CHINA SECURITIES REGULATORY COMMISSION | | | 请输入关键字 | | | | --- | --- | --- | --- | --- | --- | | △ 首页 价 机构概况 ■ 政务信息 吕. 办事服务 | 同 新闻发布 | 同 互动交流 | | 00 统计信息 | 三 专题专栏 | | √ 当前位置：首页 > 政务信息 > 政府信息公开 > 主动公开目录 > 按主题查看 > 国际合作 > 境外证券发行 > 结果公示 | | | | | | | 紧 引 号 bm56000001/2025-00014859 | | ਨੇ 类 | 结果公示,备案管理 | | | | 发布机构 | | 发文日期 | 2025年12月22日 | | | | 名 称 关于北京智谱华章科技股份有限公司境外发行上市及境内 ...

但当你真正翻开这份招股书，你会发现这种"穷"，其实是一场对行业惯性的人效挑战。 效率的胜利：1%的资金与 29岁的团队 在中国的大模型牌桌上，MiniMax 始终是一个难以被归类的"另类"。 当同行们动辄融资百亿、卷入算力军备竞赛，甚至为了一个投流渠道挥金如土时，MiniMax 在招股书中甩出了一个让业界集体沉默的数据：从 2022 年成 立至今，一共才花了约 5 亿美元。 5 亿美元是什么概念？ 放在硅谷，这甚至不够 OpenAI 塞牙缝——后者的累计花销估算已达 400 亿至 550 亿美元。放在国内，这笔钱可能也就够大厂买个半年的流量包。 对这份招股书，质疑声随之而来：在 AGI 这场动辄千亿起跳的豪赌局里，区区 5 亿美元能买到通往未来的门票吗？MiniMax 是不是没钱了？是不是在这 场残酷的淘汰赛中，为了"活下去"而不得不选择了"消费降级"？ 在MiniMax 这份招股书里，硅基君最大的一个感受就是，极致的效率。 MiniMax从2023年开始进行商业化，营收已达到346万美元，2024年直接飙升到3052万美元，同比暴涨了782.2%。2025年前9个月，公司的营收额再度大涨 175%，达到53 ...

Core Insights - Tongyi released a new end-to-end voice interaction model called Fun-Audio-Chat, which emphasizes "voice-to-voice" interaction capabilities, allowing users to engage in multi-turn conversations directly through voice [1] - The model achieved leading performance in various speech and multimodal evaluations, surpassing several other models of similar parameter scale, indicating its strong capabilities in speech understanding, generation, and dialogue collaboration [1][2] Model Features - Fun-Audio-Chat-8B is part of the Tongyi Bailing voice model family, which previously included speech-to-text and text-to-speech models. Unlike its predecessors, this model focuses on end-to-end voice interaction for applications such as voice chatting, emotional companionship, smart terminal interaction, and voice customer service [1] - The model employs a two-stage training strategy called Core-Cocktail, which integrates speech and multimodal capabilities while fine-tuning existing language model parameters to mitigate the "catastrophic forgetting" issue [2] - It also incorporates multi-stage, multi-task preference alignment training to enhance the model's ability to accurately capture semantic and emotional cues in real voice conversations, improving the naturalness of dialogue [2] Efficiency and Practicality - Fun-Audio-Chat-8B features a dual-resolution end-to-end architecture that compresses, autoregresses, and decompresses audio, reducing the audio frame rate to approximately 5Hz. This design saves nearly 50% of GPU computing costs while maintaining speech quality, which is significant given the high computational costs associated with current speech models [2] - The open-sourcing of Fun-Audio-Chat-8B signifies a move towards practical applications of large speech models in real-world scenarios, emphasizing low computational power and strong dialogue capabilities [2]

头图来源|AI制图 Minimax和智谱营造了一个营业收入快速增长的氛围，但实打实的现金亏损总会敲响每一位投资者的警钟。 智谱与MiniMax在48小时内相继递表，向资本市场展示了靠开源大模型、靠产品这两个截然不同的方式获取营收高速增长的亮眼故事。 然而，光鲜之下，两家公司的增长逻辑都显得根基不牢，透着一丝"虚张声势"。 智谱主要向G端及B端客户提供模型服务。在招股书中，智谱更是多次强调自家的模型为超过八千家机构客户提供支持。智谱的收入快速增长，2022年至 2024年，智谱的收入分别为5740万元、1.25亿元和3.12亿元，年均复合增长率超过130%。 同样是重点布局B端客户，对比OpenAI与多邻国的合作，智谱的合作显得十分的"虚"。OpenAI在2021年就开始和多邻国合作，将大模型应用于测试业 务，如今多邻国更是位列OpenAI的30家Tokens消耗破万亿的客户名单首位。 首先，智谱的合作多为"次抛"，合作关系并不牢固。从2022年、2023年、2024年到2025年上半年，智谱的前五大客户完全不重合。为此，智谱在招股书中 使用了20个完全不同的字母进行区分，如果报告期再长一些，字母表恐怕都不够用 ...

Group 1 - Alibaba-W recorded a net buy of 1.36 billion, while Meituan-W saw a net buy of 0.223 billion, and Zijin Mining had a net buy of 0.111 billion [1][3] - China Mobile experienced a net sell of 1.975 billion, Tencent Holdings had a net sell of 1.088 billion, Kuaishou-W saw a net sell of 0.174 billion, and SMIC had a net sell of 0.141 billion [1][3] - Southbound funds have continuously net sold China Mobile for 8 days, totaling 6.49287 billion HKD [3] Group 2 - Alibaba has released a new end-to-end voice interaction model called Fun-Audio-Chat, which includes open-sourced 8B model weights and inference code [4] - Zijin Mining benefits from the rise in gold prices, which have increased over 71% this year, driven by expectations of interest rate cuts by the Federal Reserve and geopolitical risks [4] - Ganfeng Lithium reports a gradual increase in lithium carbonate futures prices, with optimistic market expectations for future lithium prices due to recovering quarterly performance in lithium companies [4] Group 3 - Kuaishou announced that its live streaming function was attacked on December 22, 2025, but has since been restored, with other services unaffected [5] - Shijiazhuang Pharmaceutical Group's shareholder Cai Dongchen increased his stake by acquiring 3.454 million shares at an average price of 8.1957 HKD per share, totaling approximately 0.11 billion HKD [5]

Core Insights - Mianbi Intelligent has recently completed a financing round of several hundred million yuan, with participation from various investors including Jingguorui, Guoketouzi, CICC Porsche Fund, Mijucapital, and Heji Investment [1][2] - The funds raised will primarily be used to enhance research and development of efficient edge large models and accelerate the commercialization of edge AI [1][2] Company Developments - The successful completion of this financing round is attributed to the expanding market space for edge intelligence and the investors' recognition of Mianbi's technological strength, market position, and industry prospects [2] - Mianbi's MiniCPM edge model has achieved large-scale deployment across multiple sectors, including automotive, mobile, PC, and smart home, collaborating with well-known companies such as Geely, Changan, Volkswagen, and Huawei [2] Industry Context - Mianbi's CEO, Li Dahai, emphasized the commitment to partnering with industry chain collaborators to enable efficient edge large models to operate on a vast number of terminals, providing innovative and inclusive smart experiences for consumers [2] - The company is positioned to leverage favorable national policies, accelerated technological iterations, and emerging consumer scenarios, aiming to strengthen its leadership in the edge AI market following this financing [2]

Core Insights - The emergence of open-source foundational models like DeepSeek and Qwen has shifted the focus of the industry from expensive pre-training to a "post-training" model, significantly reducing the iteration cycle for financial vertical models from months to weeks and lowering computational requirements from "ten thousand cards" to "hundred cards," resulting in a hundredfold decrease in training costs [1][7][15]. Group 1: AI Implementation in Finance - The application of AI in serious industries like finance requires a focus on rigor, professionalism, and compliance [3][8][17]. - A "trustworthy intelligent agent" in finance relies on three pillars: a financial model as the brain, a financial knowledge base for experience, and a financial toolset for execution [3][20][21]. - The introduction of large models has revolutionized natural language understanding, significantly lowering the barriers for human-computer interaction [4][14]. Group 2: Challenges and Solutions - The financial industry faces six major pain points in implementing large models: limited computational power, insufficient and low-quality data, rapid model iteration, lack of knowledge accumulation, absence of application methodologies, and talent shortages [16]. - To address these challenges, a robust system to suppress "hallucinations" in large models is essential, as these hallucinations can increase with enhanced reasoning capabilities [3][5][17]. Group 3: Training Methodology and Future Outlook - The training of financial models should adopt a two-phase approach, balancing general and financial data to enhance capabilities without compromising general knowledge [23]. - Continuous evaluation and iteration of intelligent agents are necessary, treating their development as an ongoing process rather than a one-time software delivery [6][23]. - The application of large models in industries is not just a technological transformation but also a strategic business reshaping, necessitating a departure from traditional workflows [9][10][24].