Group 1: Market Performance - The technology sector, particularly AI hardware, led the market rally, with significant gains in the ChiNext and STAR Market indices [2] - The Shanghai Composite Index closed at 4134.02 points, up 0.05%, while the Shenzhen Component Index rose by 0.86% to 14283.00 points [2] - Total trading volume in the Shanghai and Shenzhen markets reached 21,418 billion yuan, an increase of 1,575 billion yuan from the previous day [2] Group 2: AI Hardware Sector - The focus of investment shifted to upstream hardware in the AI sector, with stocks like Unisplendour, Huanxun Electronics, and Leo Group hitting the daily limit [3] - CPO industry leader Tianfu Communication saw a 14.38% increase in stock price, reaching a market capitalization of over 250 billion yuan [3] - Vertiv, a liquid cooling supplier for NVIDIA, reported a 252% year-on-year increase in order volume, leading to a 24.49% jump in its stock price [3] Group 3: Power Equipment Sector - The power equipment sector gained attention due to increased energy demand from computing centers, with the Shenyuan Power Equipment Index rising by 3.18% [4] - Companies like Hancable and Sifang Co. saw their stock prices hit the daily limit, while major firms like Siyuan Electric and TBEA rose over 4% [4] - China's transformer export value reached a record 64.6 billion yuan in 2025, a nearly 36% increase from the previous year [4] Group 4: Market Trends and Analysis - A rapid rotation among A-share sectors has been observed, with AI, non-ferrous metals, and consumer sectors showing performance, although sustainability is weak [6] - Analysts suggest that the recent sector rotation is driven by funds seeking more price-elastic industries, indicating a potential nearing of the rotation's end [6] - The stable free cash flow generation capability of A-shares is expected to reduce reliance on financing and short-term economic cycles, enhancing shareholder returns [6]

Core Insights - The article highlights the significant advancements in China's AI models, particularly following the introduction of DeepSeek-R1, which has been recognized as a major breakthrough in the AI field, rivaling top models from American tech giants [1][5] - Chinese tech companies continue to innovate, launching powerful new AI models that are cost-effective compared to their American counterparts, drawing global attention [1][5] Group 1: AI Model Developments - The new video generation model Seedance 2.0 by ByteDance can understand complex natural language instructions and produce high-quality video content, marking a significant improvement over earlier AI video tools [3] - Seedance 2.0's ability to generate coherent multi-scene narratives from a single instruction is seen as a groundbreaking innovation in the industry [3] - Other companies, such as Zhiyuan and Alibaba, are also releasing new models like GLM-5 and Qwen 3.5, indicating a competitive landscape in AI development [4] Group 2: Market Expectations and Trends - Analysts express high expectations for the performance of new Chinese AI models, especially in light of the successful launch of DeepSeek, which has challenged previous assumptions about the necessity of extensive resources for developing advanced AI [5] - The success of DeepSeek, an open-source model, has shifted perceptions in the industry, leading more Chinese tech companies to adopt open-source approaches, thereby lowering barriers to accessing cutting-edge AI technology [5] - The performance gap between Chinese and American large language models is rapidly narrowing, suggesting a significant shift in the competitive dynamics of the AI market [5]

Group 1 - Zhipu released the open-source GLM-5 model with a parameter scale expanded to 744 billion (activated 40 billion), ranking fourth globally in the Artificial Analysis leaderboard and first in open-source, with coding and agent capabilities approaching Claude Opus 4.5 [1] - The model achieved scores of 77.8 and 56.2 in SWE-bench-Verified and Terminal Bench 2.0, respectively, setting new open-source SOTA records, excelling in complex systems engineering and long-range agent tasks [1] - GLM-5 has been adapted to domestic chips such as Huawei Ascend, Cambricon, and Kunlun, and introduced Z Code full-process programming tools and AutoGLM universal agent assistant [1] Group 2 - MiniMax launched the M2.5 model with only 10 billion activated parameters, achieving flagship-level reasoning speed three times faster than Opus [2] - The model completed a full-stack learning website in 9 minutes and can independently perform physical simulations and enterprise-level CMS system setups, supporting cross-platform development for PC/App/React Native [2] - It utilizes a native agent RL training framework and CISPO algorithm, achieving approximately 40 times training acceleration and is compatible with mainstream development tools like Claude Code and OpenClaw [2] Group 3 - Xiaohongshu's foundational model team released the open-source FireRed-Image-Edit, achieving SOTA in multiple authoritative rankings such as ImgEdit and GEdit, with code and technical reports now available [3] - The model employs a three-stage training process to enhance capabilities and innovatively introduces Layout-Aware OCR-based Reward, significantly improving text editing accuracy and style retention [3] - It supports various complex editing scenarios, including instruction-following consistency, text editing, style transfer, multi-image fusion, and old photo restoration, with model weights set to be open-sourced [3] Group 4 - Xiaomi released the open-source VLA model Xiaomi-Robotics-0 with 4.7 billion parameters, excelling in visual language understanding and real-time execution capabilities, achieving optimal results in comparisons across 30 models including LIBERO, CALVIN, and SimplerEnv [4] - The model uses a Mixture-of-Transformers architecture, where the VLM brain understands instructions and the Diffusion Transformer generates high-frequency smooth actions [4] - It addresses action discontinuity issues through asynchronous reasoning and Λ-shape attention masks, enabling real-time inference on consumer-grade graphics cards, and has been open-sourced on GitHub and HuggingFace [4] Group 5 - Gaode launched the ABot series of embodied base models, with ABot-M0 responsible for operations and ABot-N0 for navigation, achieving comprehensive SOTA across 10 global authoritative evaluations [5][6] - ABot-M0 integrates 6 million cross-platform trajectory data through action language and proposes an action manifold learning algorithm, achieving an 80.5% success rate on Libero-Plus, surpassing pi0 by nearly 30% [6] - ABot-N0 unifies five core navigation tasks within a single VLA architecture, constructing 8,000 high-fidelity 3D scenes and 17 million expert examples, with a 40.5% improvement in SocNav success rate [6] Group 6 - Rokid Glasses launched the "customizable agent" feature on the Lingzhu platform, allowing integration with OpenClaw or privately deployed models like DeepSeek R1 and Qwen3 through a standard SSE interface [7] - Users can achieve local closed-loop processing of private data and switch model bases with one click, leveraging the ClawHub skill ecosystem to execute capabilities like file systems, browsers, and IM messaging [7] - The platform empowers users by allowing them to summon private agents via voice commands or shortcuts, creating a 24/7 intelligent assistant [7] Group 7 - Google DeepMind released the AI mathematician Aletheia based on Gemini Deep Think, achieving a score of 91.9% on IMO-ProofBench, setting a new SOTA and capable of independently writing and publishing academic papers [8] - Aletheia systematically evaluated 700 open problems in the Erdős conjecture database and autonomously solved 4 unsolved mysteries, demonstrating self-correction and acknowledgment of limitations [8] - Gemini Deep Think collaborated with experts to tackle 18 long-stagnant research challenges, resolving a decade-long submodel optimization conjecture, with one paper accepted by ICLR 2026 [8] Group 8 - HyperWrite's CEO published an article that garnered 70 million views, stating that the release of GPT-5.3-Codex and Claude Opus 4.6 marks a qualitative change in AI [9] - AI can now independently complete the workload of human experts in 5 hours, with this capability doubling every 4-7 months, and GPT-5.3 plays a crucial role in its self-training process, initiating a recursive self-improvement cycle [9] - Almost all cognitive work performed in front of screens will be affected, and it is advised to spend one hour daily experimenting with AI, as the current cognitive window period will not last long [9] Group 9 - Anthropic released a 53-page report warning that the risks associated with Claude Opus 4.6 are approaching ASL-4 levels, outlining 8 potential risk pathways that could lead to catastrophic harm, including autonomous escape and autonomous operation [10][11] - The report concludes that current models do not exhibit "sustained consistent malicious intent," and the risk of catastrophic damage is "very low but not zero," entering a "gray area" of capability assessment [10] - The head of Anthropic's safety research team resigned, stating that "the world is in crisis," and xAI co-founder predicts that recursive self-improvement cycles may be launched within 12 months [11]

Core Viewpoint - The article discusses the competitive landscape of AI model releases during the Chinese New Year period, highlighting the shift from model performance to efficiency and practical applications, with DeepSeek's V4 model being a focal point for potential industry transformation [2][4][15]. Group 1: Industry Dynamics - The 2026 Chinese New Year is expected to see a surge in flagship model releases, with multiple companies, including ByteDance, Alibaba, and DeepSeek, preparing significant updates [4][5][8]. - ByteDance has already launched three models, signaling a strong market entry with its Seedance 2.0 model [4]. - Alibaba plans to release Qwen 3.5 in mid-February, supported by a substantial customer acquisition incentive [5]. - DeepSeek's V4 model is anticipated to enhance coding and long prompt processing capabilities, with a reported support for up to 1 million tokens [8][15]. Group 2: Competitive Implications - The simultaneous release of multiple models may lead to a "winner-takes-all" scenario, where underperforming models face significant disadvantages [9][10]. - The Chinese New Year period is characterized by scarce attention, making it crucial for companies to present credible flagship updates to remain relevant [12][13]. Group 3: DeepSeek's Strategic Focus - DeepSeek's potential release aims to improve efficiency through a novel "conditional memory" approach, which could shift expensive computations to more cost-effective retrieval operations [15][16]. - If successful, this could enable AI to transition from being an expensive "toy" to a cost-effective "tool," facilitating its integration into high-frequency consumer products [17]. Group 4: Beneficiaries of the Model War - Tencent is positioned as a major beneficiary of the upcoming model releases, leveraging its high-frequency communication platforms like WeChat and QQ to enhance user experience with improved AI capabilities [19][20]. - In contrast, Alibaba and Baidu may face a dual challenge: while stronger models can enhance user experience, a price war initiated by DeepSeek could pressure the entire API service market [21][22]. Group 5: Market Sentiment and Future Outlook - Despite the excitement in the capital markets, there is skepticism regarding the actual performance of consumer-facing AI models, with large-scale user testing during the Chinese New Year serving as a critical evaluation point [24][25]. - The true signal for adoption will be whether major players integrate AI as a default feature in high-frequency interfaces, which would drive sustained demand for AI capabilities [25]. Group 6: Valuation Perspectives - Morgan Stanley maintains a bullish outlook on model developers like Zhiyu and MiniMax, projecting significant long-term growth based on their advancements and market positioning [27][28]. - The valuation logic is shifting towards long-term profitability, with target prices set at 400 HKD for Zhiyu and 700 HKD for MiniMax, based on expected earnings by 2030 [29][30].

这款神秘模型究竟是何方神圣？在前几天，它引发了一大波猜测，在马年春节来临前它以"小马"命名， 很多人认为它来自中国的实验室，最后猜测的结果集中在智谱的GLM-5和DeepSeek的DeepSeek-V4上。 现在，GLM-5终于正式露面，其参数规模从355B（激活32B）扩展至744B（激活40B）。在Artificial Analysis榜单中，GLM-5 位居全球第四、开源第一。 此外，GLM-5已完成与华为昇腾、摩尔线程、寒武纪、昆仑芯、沐曦、燧原、海光等算力平台的深度 推理适配。智谱方面表示，通过底层算子优化与硬件加速，GLM-5在国产芯片集群上已经实现高吞 吐、低延迟的稳定运行。 在国内大厂掀起AI入口之争的同时，另一场AI Coding（AI编程）的大战也在全球AI赛场上展开。当地 时间2月5日，Anthropic发布Claude Opus 4.6，编码技能大幅提升。几乎是与其同步，OpenAI紧急发布了 GPT-5.3-Codex。一场围绕企业级软件开发市场的争夺战，正式打响。紧接着Pony Alpha的上线，无异 于是对海外市场的"突袭"，也是国产模型在AI Coding这个赛场上一次非常重要 ...

Group 1 - The core announcement is that Moole Technology (688795.SH), known as the "first domestic GPU stock," has completed the full adaptation and verification of the new generation large model GLM-5 from Zhipu AI (02513.HK) on its flagship AI training and inference GPU MTT S5000 using the SGLang inference framework [2] - GLM-5 is positioned as a top-tier coding model with a 20% performance improvement over its predecessor, featuring breakthroughs in Agentic Engineering capabilities, enabling end-to-end development from requirements to applications [2] - GLM-5 ranks fourth globally and first in open-source on the authoritative Artificial Analysis list, achieving state-of-the-art scores in mainstream benchmark tests, surpassing Gemini 3 Pro [3] Group 2 - The MTT S5000 GPU is designed specifically for large model training, inference, and high-performance computing, with a maximum AI computing power of 1000 TFLOPS, 80GB of memory, and a memory bandwidth of 1.6TB/s [3] - The inter-card bandwidth of the MTT S5000 is 784GB/s, and it fully supports precision calculations from FP8 to FP64 [3]

Core Insights - The article highlights the significant advancements in China's AI models, particularly the DeepSeek-R1, which has achieved performance comparable to top US models at a much lower cost and computational power, marking a pivotal moment in the AI industry referred to as the "DeepSeek moment" [1][6] - Chinese tech companies continue to innovate and release powerful new AI models, maintaining momentum in the industry despite global challenges [6][10] Group 1: AI Model Developments - ByteDance has launched the Seedance 2.0 video generation model, capable of understanding complex natural language instructions and producing high-quality video content, representing a significant leap from earlier AI video tools [3][9] - The Seedance 2.0's groundbreaking feature is its ability to generate coherent multi-scene narratives from a single instruction, maintaining consistency in characters, lighting, style, and atmosphere, achieving near "zero post-production" results [3][9] - Other companies, such as Zhiyuan and Alibaba, are also making strides with new models like GLM-5 and the upcoming Qwen 3.5, respectively, indicating a competitive landscape in AI development [4][10] Group 2: Market Perception and Trends - Analysts note that the reasoning capabilities of Chinese AI models are on par with their US counterparts, but at significantly lower deployment costs, drawing global attention [6][10] - The success of DeepSeek has shifted perceptions regarding the necessity of massive investments in computational infrastructure for developing advanced AI, challenging the notion that only companies spending billions can create cutting-edge AI [5][10] - The trend towards open-source AI models, exemplified by DeepSeek, is lowering barriers for developers and businesses to access advanced AI technologies, making "open-source and low-cost" the new norm in the industry [5][10]

Group 1 - Zhao Ming, former CEO of Honor, will join Qianli Technology's board and may serve as co-chairman, focusing on AI business model strategies [1] - Qianli Technology, previously known as Lifan Technology, is led by Yin Qi and is concentrating on "AI + automotive" to create a global smart mobility technology brand [2] - The Qianwen App reported that in just six days, users made 41 billion requests to "Qianwen help me," resulting in over 120 million AI-assisted orders, marking a significant milestone in AI task execution [6] Group 2 - Alipay's "AI Pay" has surpassed 120 million payment transactions in one week, becoming the first AI-native payment product to achieve this milestone [3] - The internet regulatory authority is intensifying efforts to address the dissemination of false information without AI identification, leading to the removal of over 540,000 pieces of illegal content [4][7] - The Seedance 2.0 video generation model has been fully launched, allowing users to create short videos through the Doubao App and other platforms [5]

Core Viewpoint - The Hong Kong IPO market has experienced a rare phenomenon of "zero first-day losses" this year, attracting significant attention from market participants [2][4]. Group 1: IPO Market Performance - A total of 22 new stocks have been listed in Hong Kong this year, with none experiencing a first-day loss, contrasting sharply with December 2025, where 10 out of 25 new listings failed to maintain their issue price on the first day [2]. - Hai Zhi Technology (02706) showed exceptional performance in dark trading, surging by 269.5% at one point and closing at 84 HKD, resulting in a substantial gain of 210.42% [2]. Group 2: Factors Influencing Market Trends - The "zero first-day loss" phenomenon is attributed to multiple factors, including positive short-term market sentiment, cautious and reasonable pricing strategies by issuers, and the stabilizing effect of cornerstone investor mechanisms [4]. - The overall quality of new listings has been high, with many being leaders in their respective niche markets, such as AI and semiconductor sectors [5]. Group 3: Key Players and Investments - Hai Zhi Technology is set to officially list on the Hong Kong Stock Exchange, having completed 13 rounds of financing since its establishment in 2013, with major institutional investors like Junlian Capital holding 13.62% [3]. - Cornerstone investors, including global sovereign funds and major asset management firms, have participated in all 22 new listings this year, contributing to a significant increase in investment amounts [6][7]. Group 4: Regulatory Environment - The Hong Kong Securities and Futures Commission has emphasized the importance of maintaining high IPO quality, addressing issues that arose during the surge of new listings in 2025 [8].

Core Insights - The Chinese AI industry is experiencing an unprecedented wave of flagship model releases, with companies racing to convert technological capabilities into scalable consumer products [1][4] Group 1: Recent Developments in AI Models - DeepSeek has emerged as a strong competitor, surpassing all major AI applications in active user numbers during the 2025 Spring Festival [1] - On February 11, Zhiyuan launched its new flagship model GLM-5, focusing on programming and agent capabilities, achieving state-of-the-art performance in coding tasks [2] - MiniMax introduced its flagship programming model MiniMax M2.5 on February 12, designed for agent scenarios and achieving high performance in various productivity applications [3] Group 2: Market Reactions and Financial Performance - Following the announcements of new models, Zhiyuan's stock price increased by 28.68% and MiniMax's by 14.62%, with market capitalizations reaching 179.229 billion HKD and 184.418 billion HKD respectively [3] - Morgan Stanley's recent report rated both Zhiyuan and MiniMax as "overweight," highlighting their potential to capture the next wave of global AI value creation [4] Group 3: Competitive Landscape and Technological Advancements - The focus of AI model development has shifted from parameter scale to practical capability upgrades, with several companies releasing or upgrading models since late January [5][6] - Alibaba's Qwen3-Max-Thinking model, launched on January 26, boasts over one trillion parameters and significant improvements in various dimensions [5] - The introduction of Kimi K2.5 by Moonlight and the updates from DeepSeek and ByteDance indicate a trend towards enhancing multi-modal capabilities and increasing model capacity [6] Group 4: Future Trends and Industry Outlook - The arrival of the agent era is reshaping the internet value chain, moving from traditional traffic scale to a focus on execution capabilities and result-based business models [7] - The new value system emphasizes cloud platforms, computational resources, governance tools, and content monetization as primary profit drivers [7]