Core Insights - MiniMax, a Shanghai-based AI company, has launched the world's first open-source large-scale hybrid architecture inference model, MiniMax-M1, which ranks second globally among open-source models [1] - The company has also released video generation model Hailuo 02, which achieved 300 million views within a week of its release on social media [1][6] - MiniMax distinguishes itself by not following mainstream dense architectures and traditional attention mechanisms, focusing instead on AGI since before the rise of ChatGPT [1][8] Performance and Cost Efficiency - The competition in large models is shifting from mere parameter scale to efficiency, cost, and overall implementation capabilities [2] - M1 supports an impressive context input of 1 million tokens, comparable to Google's latest closed-source model Gemini 2.5 Pro, while its reinforcement learning phase cost only $535,000 [2] - Hailuo 02 directly competes with Google's third-generation video generation model Veo3, showcasing superior performance in generating coherent and logical video sequences [3] Innovation in AI Video Generation - Hailuo 02 has pioneered a new category of AI video called "Animal Olympics" [4] - The development of Hailuo 02 involved collaboration with a diverse team of directors, screenwriters, and artists to ensure high-quality output [5] - High-quality data, innovative algorithms, and meticulous training processes are cited as key factors in the success of Hailuo 02 [6] Strategic Positioning - MiniMax remains one of the few startups still committed to foundational model research amidst a trend of major companies reducing their efforts in this area [7] - The company is exploring "sparse activation" MoE architecture to reduce computational costs, diverging from the prevalent dense architecture approach [8] - MiniMax aims to stay competitive in the long-term race of large model development, collaborating with other major players in Shanghai's AI ecosystem [9]

如果一个技术用3年时间，从默默无闻到改变世界，我们称其为第N次工业革命； 如果这个技术从论文到落地的过程中，头部玩家如走马灯更迭，百亿流量砸下去只听一声响就陷入沉寂，我们称其为资本绞肉机； 如果这个技术集齐了以上所有特质，还能以常规技术乘十的速度进化，让互联网时代的生态护城河、资金壁垒、规模效应全部失效，创业公司也能站在世 界舞台中心，那么，它是大模型。 而这一趋势，自今年年初DeepSeek"掀桌"以来，变得尤为明显。时至今日，在大模型行业，能留在桌子上的企业唯一的生存法则就是——创新至上。 那么，如何理解大模型行业的创新至上？为什么传统互联网的玩法在大模型时代会彻底失效？为什么大模型的淘汰，能以季度为单位发生？ 刚刚结束的MiniMax Week，或许正是我们审视这些问题的最佳切入点。 以此为切口，你会看见一个创新行业的创业公司如何挣脱巨头的射程；可以看到大模型从技术创新到改变世界之间的桥梁如何搭建；以及，一群"偏执"的 人，如何在这个一切被按下加速键的行业中破浪前行。 一只跳水猫咪，如何让世界Aha 一开始，MiniMax Week的关注度，还局限在大模型行业内的讨论：这次MiniMax会拿到多少SOT ...

Core Viewpoint - The article discusses the transformative impact of large models in the tech industry, emphasizing that innovation is the key survival strategy for companies in this space, especially in light of the rapid evolution and competition among startups and tech giants [2][3][14]. Group 1: Industry Trends - The large model industry is experiencing a significant shift towards innovation, with traditional internet business models becoming obsolete [3][4]. - The recent "Aha Moment" in the industry, exemplified by viral videos of animals performing complex actions, highlights the advancements in video AI technology and its potential [7][8]. - The MiniMax Week event serves as a critical point for examining how startups can thrive amidst competition from larger firms [4][6]. Group 2: Technological Innovations - MiniMax's Hailuo 02 model has seen a threefold increase in parameters compared to its predecessor, achieving native 1080P resolution and generating 10 seconds of high-definition content [9][10]. - The model's innovative NCR architecture allows for efficient resource allocation, significantly reducing memory read/write by over 70% and improving training and inference efficiency by 2.5 times [12][23]. - MiniMax's M1 model, with 456 billion parameters, supports the longest context length in the industry, enhancing its performance in complex tasks [16][18]. Group 3: Competitive Landscape - Despite the initial dominance of tech giants in the large model space, startups like MiniMax have captured significant market share and achieved top rankings in performance benchmarks [15][16]. - The article notes that the rapid evolution of large models requires companies to continuously innovate to maintain a competitive edge, as capital alone is insufficient for success [14][15]. - MiniMax's innovative approaches, such as the use of mixed attention mechanisms and the CISPO training method, have allowed it to outperform competitors while reducing costs [20][21][23]. Group 4: Agent Applications - The emergence of agent applications, such as MiniMax Agent, represents a new frontier in AI, enabling more complex task execution and planning capabilities [30][32]. - MiniMax Agent has been integrated into daily operations, demonstrating its effectiveness in various tasks, including programming and content creation [31][32]. - The synergy between large model innovations and agent applications is expected to drive further growth and development in the AI ecosystem [32][34].

郭晨凯 制图 这是一个"社会名声远高于商业成果"的行业。以"六小虎"（智谱、月之暗面、百川智能、MiniMax、阶 跃星辰和零一万物）为代表的中国AI大模型初创企业，在过去一年经历了融资退潮、巨头挤压与战略 分化等多重考验。 日前，证监会宣布在上交所科创板设置科创成长层，允许尚未盈利的高研发投入AI企业申请IPO。不少 创业者、投资人雀跃："这相当于是给行业打了一剂强心针。" 但也有业内人士态度审慎：短期内，IPO是一个有效解决方案；长期看，找到持续可行的商业化路径才 是破局之道。 目前，"六小虎"在商业路径上已分化为两派："转型派"不同程度地收缩了对基础模型的预训练，向小模 型过渡；"坚守派"则在咬定基座模型研发不放松的同时，探索智能体等商业化落地方向。 战略分化源自相似的困境：烧钱不止，盈利无期。其背后的艰难求索，折射出中国AI产业发展路径的 复杂性与曲折性。 在IPO的曙光前，科技光环逐渐褪去，商业大考即将到来。 ◎记者 郭成林 郑维汉 坚守与转型并存 2025年以来，曾以基座模型能力著称的大模型"六小虎"的市场关注度持续下降，不复昔日"AI明星"之 姿。 梳理"六小虎"近况，曾被视为同一梯队的六家明星 ...

证 券 研 究 报 告 计算机行业重大事项点评 MiniMax：推理模型、 （维持） Agent 与多模态 事项： ❑ 6 月 17 日，MiniMax 连续五天推出 AI 新品，涉及新的推理模型、视频生成 模型、Agent、语音设计等产品。在发布周首日开源 MiniMax-M1 模型，次日 上线海螺 AI 视频生成模型，后续发布语音设计工具，覆盖推理、视频生成、 智能体等新工具。 评论： 行业研究 计算机 2025 年 06 月 26 日 推荐 华创证券研究所 证券分析师：吴鸣远 邮箱：wumingyuan@hcyjs.com 执业编号：S0360523040001 行业基本数据 | | | 占比% | | --- | --- | --- | | 股票家数(只） | 338 | 0.04 | | 总市值(亿元） | 42,064.50 | 4.18 | | 流通市值(亿元） | 36,270.94 | 4.60 | 相对指数表现 | % | 1M | 6M | 12M | | --- | --- | --- | --- | | 绝对表现 | -2.8% | -7.9% | 35.9% | | 相对表现 | ...

本周AI商业焦点必读 本周核心趋势 2025年6月24日 1、 「功能扩展」 腾讯元宝、DeepSeek、豆包掀起AI编程新玩法 本质变革：AI将"需求→代码"链路压缩至一句话指令，传统编程工具和低代码平台遭降维打击！ 全栈覆盖 （2025.6.17-6.24） 具身智能产业化加速： 制造巨头纷纷押注具身智能机器人领域，以不断提升"机器服务密度"的 方式逐步替代传统人力配置，从而重新配置其生产业务线的人力资源结构。 多模态进入成本血拼阶段： 随着视频生成成本骤降和开源模型大量涌现，开源方案正强势重构 整个创作生态。 AI算法陪伴重构消费逻辑： "孤独经济"持续催生陪伴型消费需求，Z世代表现出强烈的付费意 愿，愿意为具有仿生记忆功能的AI陪伴服务买单。 自然语言终结传统开发模式： AI编程工具正深刻重构产业链，将复杂的"需求→代码"流程高度 压缩为直观的对话指令；低代码平台的固有价值因此被显著削弱，开发范式正加速向意图层迁 移。 交互革命 场景屠杀 原文链接： 啊？豆包居然也开始卷AI编程了？ 元宝搭载DeepSeek V3实现10+语言编程（Python/Java/C++），0配置实时运行 DeepSite ...

报告导读： GPT-5 将于 2025 年夏季发布，将整合现存模型功能。 Marvell 发布 2nm SRAM ，并上调市场预期。 MiniMax 相继推出推理模型、视频生成模型、通用智能体。 投资建议： 维持行业增持评级，推荐 AI 算力方向、云厂商方向、 AI 应用方向、 AI 社交方向以及 AI 注 入周期下的中概巨头。 GPT-5 将于 2025 年夏季发布，将整合现存模型功能。 当地时间 6 月 18 日， OpenAI 首席执行官 Sam Altman 在公司首期博客节目中探讨了通用人工智能（ AGI ）、 " 星际之门 " 计划及用户隐私等议 题。他在节目中宣布 GPT-5 预计于 2025 年夏季发布，具体日期尚未披露。该模型计划整合 OpenAI 全部旗舰功能，包括 GPT-4o 的自然语言处理能力与 o3 在代码及科学推理领域的优势，形成综合性能 更强大的新模型。 Altman 同时表示不排斥在 ChatGPT 中投放广告，此举或将成为公司拓展盈利的新渠 道。 Marvell 发布 2nm SRAM ，并上调市场预期。 在 6 月 18 日结束的 Custom AI 投资者日活动上， M ...

Domestic News - Huawei announced the open-source release of its Cangjie programming language on July 30, aimed at full-scene application development and supporting operating systems like Euler and Harmony [2] - Yushu Technology completed a C round financing of approximately 700 million RMB, achieving a post-financing valuation of about 12 billion RMB, and is actively pursuing an IPO, primarily targeting the A-share market [3][32] - Huawei launched its new generation Ascend AI cloud service based on the CloudMatrix384 super node, which integrates 384 Ascend NPUs and 192 Kunpeng CPUs, achieving a single-card inference throughput of 2300 Tokens/s [4] - Dingjie Smart released the "2025 Generative AI Enterprise Application Practical Report," analyzing ten major applications of generative AI in enterprises [5] - Ant Group established a new technology company in Hangzhou with a registered capital of 30 million RMB, focusing on software development and data processing services [6] - Huawei introduced the CloudRobo embodied intelligence platform, which integrates various capabilities for data synthesis, model development, and cloud-edge collaboration [7] - Yuxin Technology submitted its listing application to the Hong Kong Stock Exchange [8] - Zhongke Xingtou resumed participation in military material engineering service procurement activities after a suspension [9] - RIK Chemical plans to establish a joint venture for intelligent computing technology with a registered capital of 40 million RMB [10] - Juchip Technology announced the successful promotion of its new AI audio chips, achieving significant market traction [11] - The first two data center REITs in China have been approved, marking a significant expansion in the public REITs market [12] International News - Microsoft plans to lay off thousands of employees, primarily in the sales department, with the announcement expected after the end of the fiscal year [22] - OpenAI is offering discounts on its ChatGPT enterprise version, with expected annual revenue from enterprise clients nearing $15 billion by 2030 [23] - Intel appointed new executives to enhance its sales and engineering leadership [24] - South Korea plans to invest 16.1 trillion KRW in artificial intelligence over the next five years [27] - Elon Musk's AI company xAI's acquisition of X is under investigation by the EU, with potential fines expected [29] - Medical technology company Semler Scientific appointed a Bitcoin strategy director, planning to hold 105,000 Bitcoins by the end of 2027 [30] - TerraPower, founded by Bill Gates, completed a $650 million financing round with participation from Nvidia's venture capital arm [31] Policy & Trends - Over 300 cities in China have achieved 5G-A coverage, with more than 10 million users [38] - The Ministry of Education in China emphasized the need to enhance population quality in the western regions through education [39] - The National Internet Information Office reported that 433 large models have been registered in China [40] - China Mobile's chairman stated that silicon-based life forms will surpass humans in number, creating a new "population dividend" [41] - Huawei's chairman noted that the telecom market is entering a phase of demand fragmentation and technological iteration [42] - The China Securities Regulatory Commission announced support for more frontier technology companies to apply for the Sci-Tech Innovation Board [43] - Wuhan introduced measures to support talent development in the AI sector [44] - The Central Cyberspace Administration of China launched a campaign to address the misuse of AI technology, achieving significant progress in the first phase [45]

魏桢 weizhen@huayuanstock.com 证券研究报告 传媒 行业定期报告 hyzqdatemark 2025 年 06 月 23 日 证券分析师 陈良栋 SAC：S1350524100003 chenliangdong@huayuanstock.com 陈逸凡 SAC：S1350525040003 chenyifan@huayuanstock.com 腾讯上线新短剧小程序，关注 AI 生视频产业进度 投资评级： 看好（维持） ——传媒互联网行业周报（2025.6.16-2025.6.22） 投资要点： 本周观点：腾讯微信端近期上线新短剧小程序【短剧】，产品形态简洁，且采用免 费模式。此外，MiniMax 发布新视频生成模型 Hailuo 02，建议重视多模态模型迭代 中，短剧、动画等内容形态产品的生产工业化技术的突破。同时，我们建议调整中 继续重视优质新品的强势表现带来的弹性交易机会。 联系人 板块表现： 微信端上线新短剧小程序【短剧】，目前产品形态较为简洁，仅有"剧场"和"观 看历史"两个页面，其中"剧场"界面包含"口碑榜"与"推荐"两部分，"口碑 榜"主打"高传播的口碑好剧"，"推荐"主打 ...

June 18, 2025 05:12 AM GMT Kuaishou Technology | Asia Pacific AI Video Generation Competition Increases with a Strong Debut of Seedance 1.0 Pro We see a change in the competition of AI video generation with the release of two new models recently. What is new: On 11 June, ByteDance released its AI video generation model Seedance 1.0 pro in its Volcano Engine Force Conference. Seedance 1.0 now ranks No.1 on the Artificial Analysis Leaderboard for both text-to-video and image-to-video, surpassing Google's Veo ...