智通财经APP获悉，中金发布研报称，首次覆盖MINIMAX-WP(00100)给予跑赢行业评级，目标价 1,109.00港元，基于P/S估值法，对应2027年估值倍数75x。公司是目前国内少数在基础模型能力、全球 化AI原生应用商业化两端同时跑通的公司，在AI赛道中具备稀缺性。 公司将AI能力深度嵌入内部工作流;截至3Q25，公司仅385人，研发占比73.8%;CEO以下设立不超过三层 职级。AI原生、扁平、高效组织支持模型与产品的高效迭代。 该行与市场的最大不同? 中金主要观点如下： 国内最早押注原生全模态融合路线的基模厂商之一 随着Agent、世界模型推进，全模态理解与生成正逐步成为重要的基础能力。公司从成立初期即同步推 进文本、语音、视频模型研发，押注全模态融合、形成统一技术栈、积累了先发经验。 国内少有的验证海外市场规模化变现能力的基模厂商 25年前三季度，海外收入占比超73%，覆盖200+国家及地区。通过AI原生产品、开发者平台直接触达高 付费意愿用户，在欧美等成熟市场快速拓展，实现自我造血并获取真实市场反馈。 坚定执行"技术即产品"路线，"前店后厂"模式加速模型与产品的拟合迭代 公司要求模型能力直 ...

Core Insights - MiniMax is one of only four companies globally that have achieved leading positions across language, video, and audio modalities, alongside OpenAI, Google, and ByteDance [3] - The company allocates over 80% of its resources to model layers and infrastructure, emphasizing that the model itself is the core product, while applications serve as a showcase [3] - The M2 model, released in October, has become the largest domestic AI programming model in terms of real token usage, surpassing all other domestic models combined [3] - MiniMax aims to provide higher "per dollar intelligence," focusing on global collaboration to create advanced models with fewer resources [3] - The Agent AI product has surpassed the capabilities of ordinary interns in tasks such as report writing and HR functions, indicating a significant advancement in internal operations [3] Industry Dynamics - The AI large model industry differs fundamentally from traditional internet sectors, with market space driven solely by the intelligence level of models, which can unlock new markets with each significant advancement [5][11] - The industry is experiencing rapid growth, with annual revenues nearing $30 billion and a monthly growth rate of approximately 10%, indicating a highly competitive environment [19][20] - Despite the rapid growth, the number of companies capable of consistently releasing leading models is decreasing, with only about ten players remaining in the global market [20][23] - The industry is characterized by a unique closed-loop effect where each model's intelligence leap unlocks new applications, leading to increased revenue that can be reinvested into further model development [13][16] Company Strategy and Vision - MiniMax's strategy is to create a universal model that serves multiple scenarios, moving away from the previous model of needing specialized models for each new client or scenario [24][25] - The company has positioned itself as a pioneer in developing multi-modal models, integrating language, visual, and audio capabilities to achieve general artificial intelligence [25][26] - The company emphasizes that successful model development is akin to a system engineering project, requiring a cohesive and efficient research organization rather than merely accumulating resources [6][29] - MiniMax's core products include language, video, and audio models, with a focus on global commercialization and user experience driven by the underlying models [30][38] Recent Achievements - The company has made significant advancements in its models, achieving global leadership in various modalities, including a leading position in voice models and video generation [31][32] - The M2 model has quickly gained traction in the AI programming field, becoming the most widely used domestic model in this area, indicating a breakthrough for domestic AI capabilities [34] - MiniMax's video generation model, "海螺," has become one of the largest platforms globally, demonstrating the company's ability to rapidly scale and penetrate the market [32][33]

Summary of Key Points from the Conference Call Industry and Company Overview - The conference call discusses advancements in the AI industry, specifically focusing on the Deepseek V3.2 model developed by DeepMind, which showcases significant improvements in reinforcement learning and inference efficiency [1][3][5]. Core Insights and Arguments - **Model Architecture and Mechanisms**: Deepseek V3.2 introduces the Dynamic Spatial Attention (DSA) mechanism, replacing the previous Multi-Level Attention (MLA) mechanism. DSA optimizes computational efficiency by focusing on key attention parameters, particularly in complex tasks [3][5]. - **Performance Enhancements**: The C9 version of Deepseek V3.2 utilizes approximately 10% of the pre-training computational resources to significantly enhance its performance in complex tasks, such as code debugging, achieving a global leading level [1][3]. - **Context Management Strategy**: The model employs an efficient context management strategy that intelligently handles frequent task switching, multi-turn dialogues, and ambiguous inputs, effectively reducing inference costs [1][3]. - **Synthetic Data Utilization**: The training process for Deepseek V3.2 incorporates a substantial amount of high-difficulty synthetic data, which has doubled compared to previous versions. This data is crucial for the subsequent reinforcement learning phase and requires significant computational resources [1][6]. - **Open Source Innovations**: Deepseek has made strides in open-source capabilities by completing a comprehensive post-training process and supporting agent invocation, potentially leveling the playing field with closed-source models [7]. Additional Important Insights - **Reinforcement Learning Developments**: The evolution of reinforcement learning techniques has been marked by the introduction of human prompts based on Rubik's rules, enhancing the model's ability to think and execute simultaneously, thus improving overall efficiency [8][9]. - **Future of Model Pricing**: It is anticipated that by 2026, the cost of models will significantly decrease, potentially dropping to one-fifth of current prices due to advancements in technology and competitive pricing strategies among vendors [2][20]. - **Impact of Sparsity Techniques**: The implementation of sparsity techniques is expected to lower training computational requirements while increasing the upper limits of model training, encouraging more startups to engage in large model development [2][19]. - **Vertical Scene Task Solutions**: The application of reinforcement learning in e-commerce platforms illustrates the model's ability to adapt recommendations based on user feedback through multi-turn dialogue mechanisms, enhancing user satisfaction [12]. Conclusion - The advancements in Deepseek V3.2 highlight a significant shift in the AI landscape, emphasizing the importance of efficient computational mechanisms, the role of synthetic data, and the potential for open-source models to compete with proprietary solutions. The expected decrease in model costs and the rise of new startups indicate a dynamic and evolving market landscape [1][2][20].