当前内容仅限订阅用户查看 （原标题：智通决策参考︱热点围绕机器人及大模型发酵 本周留意英伟达(NVDA.US)财报） 精品VIP投研内容 VIP 立即订阅 ...

本文自南都。 采写 | 南都N视频记者 林文琪 2月21日，"全球大模型第一股"智谱AI（02513.HK）发布致歉信称，GLM-5发布后GLM Coding Plan存在透明度不够、GLM-5灰度节奏太慢、老用户升级 机制设计粗糙三大问题，表示"挨打要立正"，并提出了相应的处理和补偿方案。致歉信提到，GLM-5发布后，流量超出预期，智谱扩容节奏没有跟上，不 得已将GLM-5按照Max、Pro、Lite的顺序逐步开放。 值得一提的是，截至最新交易日（2月20日），智谱股价暴涨42.72%至725港元，市值突破3200亿港元。 | 今开 | 534.000 | | 最高 | 725.000 | | | 成交量 526.69万股 | | --- | --- | --- | --- | --- | --- | --- | --- | | 昨收 | 508.000 | | 最低 | 532.500 | | 成交额 | 32.41亿 | | 换手率 | 2.38% | | 市盈(TTM) | 亏损 | | 总市值 | 3232亿 | | / \ r _ L | Friday of the pro- | I | - | ...

Core Viewpoint - The AI company Zhiyu experienced a significant stock price surge of over 40% on February 20, but subsequently issued an apology due to operational issues following the release of its new model, GLM-5 [2]. Group 1: Apology and Issues - Zhiyu acknowledged three key mistakes in its apology letter regarding the GLM-5 release: insufficient transparency in usage rules, slow expansion pace, and a rough upgrade mechanism for existing users [6][7]. - The company implemented a tiered usage strategy to manage increased computational demands, raising peak consumption by three times and off-peak by two times, but failed to communicate this clearly to users, leading to widespread complaints [6]. - The demand for GLM-5 exceeded expectations, causing delays in service expansion and limiting access for Pro and Lite users, with Max users fully opened but facing potential throttling during peak times [6]. Group 2: User Compensation and Experience - To address user dissatisfaction, Zhiyu offered compensation, including a 15-day extension for users already utilizing GLM-5 and the option for Lite and Pro users to apply for refunds [6]. - The GLM Coding Plan, a subscription service for AI programming, saw a surge in demand due to GLM-5's strong performance, which approached state-of-the-art levels in coding capabilities [8]. Group 3: Industry Context and Future Plans - The incident reflects a broader trend in the industry, where companies like Anthropic faced similar challenges with service capacity following model releases [8]. - In response to the demand surge, Zhiyu launched a "Computing Partner" recruitment plan to enhance computational capacity, seeking partnerships with chip manufacturers and inference service providers [9][11]. - The company has already collaborated with several domestic chip platforms to optimize GLM-5 for high throughput and low latency on local computing clusters [11].

Core Viewpoint - The company, Zhiyu AI, acknowledged issues following the release of its GLM-5 model, including insufficient transparency, slow rollout, and poor upgrade mechanisms for existing users, and proposed compensation measures [2][4]. Group 1: Company Performance and Market Reaction - Zhiyu AI's stock price surged by 42.72% to 725 HKD, with a market capitalization exceeding 320 billion HKD as of February 20 [2]. - The annual recurring revenue (ARR) for the GLM Coding Plan has surpassed 100 million RMB, approximately 14 million USD [3]. Group 2: Product Launch and User Experience - The GLM-5 model was launched on February 12, achieving state-of-the-art performance in coding and agent capabilities, leading to overwhelming demand and service delays [2][3]. - Users reported increased consumption rates with GLM-5, which has over twice the parameters of GLM-4.7, designed for complex tasks [4]. Group 3: Response to User Feedback - The company issued an apology to developers, acknowledging that the demand for GLM-5 exceeded expectations, resulting in service disruptions and delays [4][9]. - A refund policy was established for affected Lite and Pro users, allowing them to apply for refunds under a "Zhiyu covers all costs" principle for the period from January 1, 2026, to February 21, 2026 [5][10]. Group 4: Future Plans and Partnerships - To address the demand-supply gap, Zhiyu AI initiated a "Computing Power Partner" recruitment plan, inviting chip manufacturers and service providers to collaborate on optimizing GLM-5's performance [3][9]. - The company plans to enhance user experience by optimizing infrastructure and combating malicious resource usage from gray market actors [12][16].

Core Insights - The company issued an apology letter acknowledging three main issues with the GLM Coding Plan: insufficient rule transparency, slow rollout of GLM-5, and a rough upgrade mechanism for existing users [2] - The GLM Coding Plan, a paid subscription service for AI programming, experienced overwhelming demand due to the strong performance of the new GLM-5 model, which achieved state-of-the-art results in key areas [2] - Following the launch of GLM-5, the company raised the prices of the GLM Coding Plan by at least 30%, and the plan sold out immediately, which is rare in the industry [2] User Experience and Demand - The high demand for the GLM Coding Plan negatively impacted user experience, with issues arising from malicious actors and resource overutilization [3] - The company is gradually opening access to GLM-5 in a tiered manner, with Max users fully open, Pro users facing potential throttling during peak times, and Lite users to be gradually opened after the holiday [4] Compensation and Future Plans - To improve user experience, the company is offering refunds for affected Lite and Pro users and extending the usage period for existing GLM-5 users by 15 days [4] - The company is actively seeking "computing partners" to enhance its infrastructure and address the high demand for GLM-5, including collaborations with chip manufacturers and service providers [5] Market Performance - On February 20, the company's stock surged by 43%, reaching a price of 725 HKD, with a cumulative increase of 524% since its listing 43 days prior, resulting in a market capitalization of 323.2 billion HKD [11]

Core Insights - The article emphasizes the importance of strategic foresight in business, highlighting that successful companies balance short-term profits with long-term goals [1][4] - Huawei is presented as a case study of a company that effectively integrates strategic thinking into its operations, particularly in its entry into the automotive industry [1][2] Group 1: Strategic Decision-Making - Companies must have a clear understanding of their strategic direction and the balance between immediate financial gains and future growth [1][4] - Effective strategic decisions are based on thorough analysis rather than impulsive actions, especially for larger enterprises where mistakes can lead to significant waste [1][2] - The ability to adapt and maintain a long-term vision is crucial for companies to avoid becoming obsolete in competitive markets [3][4] Group 2: Research and Development - The article underscores the necessity of R&D investment for maintaining competitiveness, particularly in high-tech industries [2][3] - Companies that previously relied on resource integration are now recognizing the importance of R&D, influenced by Huawei's approach [2][3] - Huawei's strategy in the automotive sector includes making its technology accessible to partners, thereby establishing a competitive edge without disrupting market rules [2][3] Group 3: Market Positioning - Companies should focus not only on sales figures but also on their market share and positioning within the industry [4] - The article warns against short-sightedness, using the example of the once-popular PHS (Personal Handy-phone System) to illustrate the risks of neglecting long-term strategy [4] - Strategic leaders must navigate the complexities of balancing short-term and long-term interests, which requires a nuanced understanding of market dynamics [4][5]

Core Viewpoint - The company, Zhipu AI, issued an apology regarding its GLM Coding Plan and announced compensation measures following the release of its flagship model GLM-5, which faced unexpected demand and operational challenges [1][10]. Group 1: Company Performance - Zhipu AI's stock price surged by 42.72% on February 20, reaching 725 HKD, with a total market capitalization of 323.2 billion HKD, marking a 524% increase from its IPO price of 116.2 HKD within 43 days of listing [10][12]. - The company has entered the top tier of market capitalization within the TMT sector in Hong Kong, surpassing competitors like Kuaishou, JD.com, Ctrip, and Baidu [5][12]. Group 2: Product Development and Pricing - The GLM-5 model, launched on February 12, demonstrated over a 20% performance improvement in programming scenarios compared to its predecessor, closely approaching the performance level of Claude Opus 4.5 [12]. - Due to high demand, Zhipu AI raised the prices of its GLM Coding Plan packages, with a 30% increase in China and over 100% for the overseas version, becoming the first domestic AI company to increase prices for large model commercial services [7][14]. - The new pricing plan sold out immediately, setting a record for the sale of paid packages for domestic AI programming models [14]. Group 3: Market Position and Future Outlook - According to a report by JPMorgan, Zhipu AI's API pricing reflects its leading capabilities and indicates a significant turning point for the company, particularly in its global API business [8][14]. - The report highlighted that the release of models GLM-4.5, GLM-4.6, and GLM-4.7, along with strategic shifts towards intelligent systems and enhanced reasoning tools, align the company's technology roadmap with global cutting-edge capabilities [8][14]. - The company is expected to see rapid increases in usage rates of GLM-5, especially in coding-centric workflows, where users exhibit high willingness to pay [8][14]. - As a newly listed stock, 60% of Zhipu AI's strategic placement shares remain locked, resulting in a limited float that amplifies stock price volatility in a capital-rich AI sector [8][14].

Core Insights - The Hong Kong stock market saw significant gains for AI companies Zhiyuan and MiniMax, with both surpassing a market capitalization of 300 billion HKD, indicating a rise of new tech leaders in the market [1][3] - On the first trading day of the Lunar New Year, while the Hang Seng Tech Index fell nearly 3%, Zhiyuan and MiniMax stood out with substantial price increases, with Zhiyuan's stock price reaching a peak of 739 HKD per share, marking a rise of over 45% [1] - MiniMax also performed well, closing at 970 HKD per share, reflecting a gain of over 14% on the same day, and a total increase of 487.88% since its listing on January 9 [3] Company Performance - Zhiyuan's market capitalization increased by over 96.7 billion HKD in one day, reaching 323.24 billion HKD, with a total increase of over 256% since February 9 [1] - MiniMax's market capitalization reached 304.23 billion HKD, with a significant rise since its IPO [3] Market Dynamics - The market landscape for tech stocks in Hong Kong has shifted, with Zhiyuan and MiniMax surpassing the market capitalizations of major companies like Ctrip, Kuaishou, and JD, and approaching the valuation of Baidu at 350 billion HKD [5] - The performance of these AI companies is seen as a reflection of the growing interest and investment in AI technologies, particularly in light of recent developments in the industry [10] Industry Trends - The rapid iteration of large models is leading to a restructuring of the internet value chain, shifting focus from traditional traffic scale to behavior execution and result-based business models [10] - The AI application sector is experiencing simultaneous technological breakthroughs and intensified commercialization, supported by national strategies promoting AI development [10] Future Outlook - Despite the strong performance of Zhiyuan and MiniMax, concerns remain regarding the overall tech sector in Hong Kong, with analysts predicting continued weakness in the market due to investor apprehension about major companies like Alibaba and Baidu [11]

Market Overview - The Hong Kong stock market indices experienced a collective decline, with the Hang Seng Index falling by 1.1% to 26,413.35 points, the Hang Seng Tech Index down by 0.9%, and the National Enterprises Index decreasing by 1.55% [1][8] - Technology stocks saw widespread losses, with Baidu dropping over 6%, Alibaba nearly 5%, and Tencent over 2% [1][8] Sector Performance - Despite the overall market downturn, large model and robotics concept stocks surged, with Zhihui rising nearly 43%, MINIMAX increasing over 14%, and both companies' market capitalizations surpassing HKD 300 billion [1][8] - Other notable gains included Yujian up over 21%, Shoucheng Holdings nearly 12%, Sanhua Intelligent Control over 5%, and UBTECH close to 5% [1][8] Real Estate Sector - Some local real estate stocks showed strength, with Hysan Development and Swire Properties B rising over 3%, and Crown Property Trust and Hang Lung Properties increasing over 2% [2][9] - Analysts from Goldman Sachs have upgraded their forecast for Hong Kong property prices from a 5% increase to a 12% increase for the year, citing favorable government visa and immigration policies, a 20% rise in rents over the past three years, and lower mortgage rates encouraging residents to consider buying instead of renting [2][9] Robotics and AI Sector - The robotics sector saw significant growth, with Yujian rising over 19%. The recent CCTV Spring Festival Gala showcased various humanoid robot startups, contributing to heightened interest in the sector [3][10] - Douyin e-commerce reported a staggering 1680% year-on-year increase in GMV for robot products from February 16 to 18, with order volume up 655% [5][12] - AI application stocks also performed well, with Zhihui increasing over 19% following the announcement of a "computing power partner" recruitment plan aimed at optimizing their GLM-5 model [6][14]

Core Insights - Google has officially launched its new flagship model, Gemini 3.1 Pro, which outperforms previous models in 12 benchmark tests, including Gemini 3 Pro, Claude Opus 4.6, and GPT-5.2, achieving the top position [1][29]. Benchmark Performance - In the benchmark tests, Gemini 3.1 Pro scored 44.4% in the "Humanity's Last Exam" academic reasoning test, surpassing Gemini 3 Pro's 37.5% [2]. - The model achieved a remarkable 77.1% in the ARC-AGI-2 abstract reasoning puzzles, doubling the performance of Gemini 3 Pro [2][29]. - For scientific knowledge, Gemini 3.1 Pro scored 94.3% in the GPQA Diamond test, indicating strong performance in factual accuracy [2]. Model Enhancements - The enhancements in Gemini 3.1 Pro focus on complex task processing capabilities, including advanced reasoning, multimodal understanding, and project generation [10][29]. - The model's ability to generate detailed SVG animations and interactive designs has significantly improved, showcasing its capability in creative programming and interactive design [21][23]. User Access and Pricing - Starting today, Google AI Pro and Ultra subscribers can access Gemini 3.1 Pro through various platforms, while free users can ask two questions [10]. - The API pricing for Gemini 3.1 Pro follows a tiered model, with input prices set at $2.00 per million tokens for prompts under 200,000 tokens, and $4.00 for those exceeding this limit [10][11]. Industry Trends - The AI industry is shifting from general capability comparisons to real-world complex task execution, with companies focusing on enhancing reasoning, engineering, and multimodal understanding [33]. - Google's recent rapid advancements, including the launch of Gemini 3 Deep Think and Gemini 3.1 Pro, emphasize the importance of developing models that can address complex real-world problems [33].