【解剖大盘】 今天 A 股走势相当强悍，沪指突破 4100 点，收 4120.43 点；关键是成交额再度突破 3 万亿元，为 31523 亿元，这是 A 股历史上第六次成交额超 3 万亿元，上一次发生在 2025 年9 月18 日。港股要逊色 很多，全天涨幅震荡，收盘微涨 0.32%。 这是没有办法的事情，海外资金占据较大话语权的港股更多是走国际线路，海外的一举一动才是左右行 情的关键。 1 月 6 日，丹麦、法国、德国、意大利、波兰、西班牙和英国发表联合声明，强调格陵兰岛属于其人 民，只有丹麦和格陵兰岛才能决定自身事务。针对这个问题，美国政府想了新招：考虑向格陵兰岛居民 发放现金， 以促成其加入美国；美国官员讨论过人均发放 1 万至 10 万美元的方案，总额接近 60 亿美 元。如果嫌少估计还能加码，最终来个全民公投都同意，欧洲也没话说。这招属实厉害，花点小钱不费 一兵一卒，就把格陵兰岛收入囊中。关键是还控制了北极航道。后遗症也明显，北约成员怎么看？会不 会就此解体？ 另一边，"和平意志-2026 "海上联合演习将于 1 月上中旬在南非西蒙斯敦附近海空域举行，中国、俄罗 斯、南非等金砖成员国参演。这凸显了金 ...

消息面上，1月8日，OpenAI推出 ChatGPT Health功能，可解答健康问题、链接智能设备、规划饮食运 动，功能和"蚂蚁阿福"相似。该模式集成于ChatGPT中，号称是一个"专门用于与ChatGPT进行健康相关 对话的独立空间"，预计将在未来几周内陆续向用户开放。此前，蚂蚁集团旗下AI健康助手正式更名 为"蚂蚁阿福"，其链接了全国5000家医院、30万真人医生在线问诊，支持打通苹果、华为、OPPO、 vivo、鱼跃等十大品牌智能设备。 2025年3月，讯飞医疗科技正式发布基于深度推理技术的"讯飞星火医疗大模型X1"。该模型是当前唯一 完全采用全国产算力训练完成的医疗深度推理大模型。在真实场景测试中，其全科辅助诊断合理率达到 94.0%。2025年7月，讯飞星火医疗大模型完成能力升级，多项性能指标得到进一步提升。西南证券认 为，AI+医疗是明确的政策增量方向，公司作为龙头充分受益。此外，公司具备自研基座大模型，性能 上限持续抬升，建议持续关注。 讯飞医疗科技(02506)涨超10%，截至发稿，涨10.01%，报89港元，成交额5394.89万港元。 ...

2025年3月，讯飞医疗科技正式发布基于深度推理技术的"讯飞星火医疗大模型X1"。该模型是当前唯一 完全采用全国产算力训练完成的医疗深度推理大模型。在真实场景测试中，其全科辅助诊断合理率达到 94.0%。2025年7月，讯飞星火医疗大模型完成能力升级，多项性能指标得到进一步提升。西南证券认 为，AI+医疗是明确的政策增量方向，公司作为龙头充分受益。此外，公司具备自研基座大模型，性能 上限持续抬升，建议持续关注。 智通财经APP获悉，讯飞医疗科技(02506)涨超10%，截至发稿，涨10.01%，报89港元，成交额5394.89 万港元。 消息面上，1月8日，OpenAI推出 ChatGPT Health功能，可解答健康问题、链接智能设备、规划饮食运 动，功能和"蚂蚁阿福"相似。该模式集成于ChatGPT中，号称是一个"专门用于与ChatGPT进行健康相关 对话的独立空间"，预计将在未来几周内陆续向用户开放。此前，蚂蚁集团旗下AI健康助手正式更名 为"蚂蚁阿福"，其链接了全国5000家医院、30万真人医生在线问诊，支持打通苹果、华为、OPPO、 vivo、鱼跃等十大品牌智能设备。 ...

重要股东主动延长禁售期，彰显长期信心 主动延锁反映了主要利益相关方对公司未来发展的高度认可，也间接确认了企业AI医疗战略及中长期 成长空间。第二，核心股东合计所持股份占比极高，禁售期延续在一定程度上稳定了市场增长预期，有 助于遏制短期股票供给压力，对股价有正面支持效应。高管及核心技术团队通过员工持股平台深度绑定 股权，充分调动团队积极性，有利于巩固业务创新动能。 深度绑定自研"星火"大模型，具备全栈自主可控的技术壁垒 2025年3月，公司正式发布基于深度推理技术的"讯飞星火医疗大模型X1"。该模型是当前唯一完全采用 全国产算力训练完成的医疗深度推理大模型。在真实场景测试中，其全科辅助诊断合理率达到94.0%。 2025年7月，讯飞星火医疗大模型完成能力升级，多项性能指标得到进一步提升。具体而言，全科辅助 诊断合理率提升至95.0%，体检报告解读合理率提升至86.3%，健康咨询解答率提升至91.5%，院端心血 管内科主要诊断合理率提升至91.2%。测试表明，在关键医疗场景任务上，其表现继续大幅领先于 OpenAI o3及DeepSeek R1等国际主流模型。与多数依赖第三方或开源模型的AI应用公司不同，讯飞医疗 ...

西南证券发布研报称，AI医疗市场空间较大，讯飞医疗科技(02506)具备自研基座大模型，性能上限持 续抬升，公司作为行业内覆盖最广的AI医疗公司，是承接政策红利和国家级项目的最主要标的，建议 持续关注。 西南证券主要观点如下： 事件 20251年12月29日公司发布有关股东自愿延长禁售期的通知，讯飞医疗科技的主要股东，包括母公司科 大讯飞（002230）股份有限公司、科讯创投等核心投资者，以及高管合伙平台与重要个人持股人，主动 将原定于2025年12月29日届满的股份禁售期再度延长一年，至2026年12月29日止，承诺期间内不减持公 司H股或非上市股份。截至公告日，上述股东合计持有公司H股52,758,395股，占H股总股本68.25%，合 计总股本持有比例达72.86%。 深度绑定自研"星火"大模型，具备全栈自主可控的技术壁垒 2025年3月，公司正式发布基于深度推理技术的"讯飞星火医疗大模型X1"。该模型是当前唯一完全采用 全国产算力训练完成的医疗深度推理大模型。在真实场景测试中，其全科辅助诊断合理率达到94.0%。 2025年7月，讯飞星火医疗大模型完成能力升级，多项性能指标得到进一步提升。具体而言，全 ...

西南证券主要观点如下： 智通财经APP获悉，西南证券发布研报称，中国医疗人工智能行业空间较大，政策推动行业快速发展； 讯飞医疗科技(02506)具备自研基座大模型，性能上限持续抬升。星火医疗大模型X1能够显著降低幻觉 现象，提升推理的正确性、专业性与可解释性；基于大模型和大数据技术平台，构建多元化GBC客户 产品矩阵。首次覆盖，建议关注。 2025年3月，公司正式发布基于深度推理技术的"讯飞星火医疗大模型X1"。该模型是当前唯一完全采用 全国产算力训练完成的医疗深度推理大模型。在真实场景测试中，其全科辅助诊断合理率达到94.0%。 2025年7月，讯飞星火医疗大模型完成能力升级，多项性能指标得到进一步提升。具体而言，全科辅助 诊断合理率提升至95.0%，体检报告解读合理率提升至86.3%，健康咨询解答率提升至91.5%，院端心血 管内科主要诊断合理率提升至91.2%。测试表明，在关键医疗场景任务上，其表现继续大幅领先于 OpenAI o3及DeepSeek R1等国际主流模型。 基于大模型和大数据技术平台，构建多元化GBC客户产品矩阵 截至2024年底，公司已向超过7万家基层医疗机构提供产品和服务，业务覆盖全 ...

Core Viewpoint - The company, iFlytek Medical, aims to demonstrate its investment value through its performance, despite the overall market uptrend in the Hang Seng Index and popular concept stocks [1] Financial Performance - For the first half of 2025, iFlytek Medical reported a revenue of 299 million yuan, representing a year-on-year growth of 30%, continuing a compound annual growth rate of 24.7% from 2022 to 2024 [1] - The company's gross margin remained stable at a high level of 51.5%, while the net loss rate significantly narrowed to 22%, improving by 15.5% year-on-year [1] Business Segments - Government segment revenue reached 140 million yuan, up 87% year-on-year, accounting for 47.4% of total revenue, with grassroots solutions growing by 52% and regional solutions by 178% [2] - The combined revenue from hospitals and patients was 156 million yuan, a year-on-year increase of 3.7%, making up 52.6% of total revenue [2] Product Matrix and Market Position - iFlytek Medical has developed a diversified product matrix for GBC (Government, Business, Consumer) scenarios, with its AI medical products leading the industry and achieving large-scale applications [2][3] - The core product "Smart Medical Assistant" has been implemented in over 69,700 grassroots institutions, completing over 1.01 billion AI-assisted diagnoses [3] Research and Development - The company invested 134 million yuan in R&D in the first half of 2025, accounting for 45% of its revenue, which has led to significant contributions from R&D outcomes [7] - iFlytek Medical's AI model, Spark Medical Model X1, has outperformed competitors in various medical tasks and has been recognized as a leader in the industry [5][7] Market Outlook - The AI medical market in China is projected to grow at a compound annual growth rate of 33.8% from 2019 to 2023, potentially reaching 315.7 billion yuan by 2033 [7] - The company is positioned to benefit from both high industry growth and supportive policies, with significant interest from institutional investors and a strong performance in the stock market [8]

Core Insights - The company reported a significant increase in revenue and gross profit for the six months ending June 30, 2025, with revenue reaching 299 million RMB, a year-on-year increase of 30.26% [1] - The growth in revenue is primarily attributed to the increase in revenue from grassroots solutions (up 52.3%), patient management services (up 10.1%), and regional solutions (up 178.1%) [1] Financial Performance - Revenue for the period was 299 million RMB, reflecting a 30.26% increase year-on-year [1] - Gross profit amounted to 154 million RMB, showing a year-on-year increase of 26.96% [1] - Research and development expenses totaled 115 million RMB [1] Product Development - In March 2025, the company launched the X1 medical deep reasoning model, which is the only medical deep reasoning model trained using fully domestic computing power [2] - The X1 model achieved a general auxiliary diagnosis accuracy rate of 94.0%, a specialist auxiliary diagnosis accuracy rate of 90.1%, and a health consultation response rate of 89.0%, all surpassing competitors like GPT-4o and DeepSeek R1 [2] Strategic Partnerships - The company partnered with Huawei to introduce a fully domestic medical integrated machine that supports the X1 model and various open-source models, enabling collaborative and rapid deployment [2] - The integrated machine includes a "one-stop" Starfire medical intelligent platform, offering 20 medical professional intelligent agents for tasks such as automatic medical record generation and report interpretation [2] Performance Enhancements - The integration of deep thinking and industry understanding has improved the medical knowledge capability by 30.0% and reduced model hallucination by 10.0% [2] - Targeted optimizations based on Huawei's Ascend hardware system enhanced the inference performance of the X1 model by 40.0% [2]

Group 1 - Company reported a revenue of 299 million RMB for the six months ending June 30, 2025, representing a year-on-year increase of 30.26% [1] - Gross profit reached 154 million RMB, with a year-on-year increase of 26.96% [1] - Research and development expenses amounted to 115 million RMB [1] Group 2 - The revenue growth was primarily driven by grassroots solutions, which saw a year-on-year increase of 52.3%, patient management services with a 10.1% increase, and regional solutions that surged by 178.1% [1] - In March 2025, the company launched the Xunfei Spark Medical Model X1, the only medical deep reasoning model trained using fully domestic computing power [2] - The model achieved a general auxiliary diagnosis accuracy rate of 94.0%, a specialist auxiliary diagnosis accuracy rate of 90.1%, and a health consultation response rate of 89.0%, all surpassing GPT-4o and DeepSeek R1 [2] - The company partnered with Huawei to introduce a fully domestic Xunfei Spark Medical Integrated Machine, which supports the X1 model and various open-source models, enhancing model collaboration and deployment [2] - The integrated machine includes a "one-stop" Spark Medical Intelligent Platform, offering 20 medical professional intelligent agents, improving medical knowledge capabilities by 30.0% and reducing model hallucination by 10.0% [2] - Targeted optimizations based on Huawei's Ascend hardware system improved the inference performance of the X1 model by 40.0% [2]

Summary of Conference Call Notes Company and Industry Overview - The conference call involved **iFlytek Medical**, a company specializing in AI-driven healthcare solutions, particularly in the field of medical big models and patient management systems [1][2]. Key Points and Arguments Financial Performance - In 2024, iFlytek Medical achieved a revenue of **734 million** (CNY), representing a **32% year-on-year growth**. Gross profit reached **404 million** (CNY), with a **28% increase** year-on-year [3][4]. - The revenue structure improved, with G-end revenue growing by **10%**, B-end (hospital services) revenue increasing by **103%**, and C-end (patient services) revenue rising by **57%** [4]. - Sales collections grew by **33%**, and operational cash flow improved by **57%** year-on-year. Accounts receivable at year-end stood at **930 million** (CNY), with **76%** from government and state-owned enterprises [5]. Technological Advancements - iFlytek Medical has developed the **X1 medical big model**, which has surpassed the capabilities of existing models like GPT-4O in medical applications. The model can achieve diagnostic accuracy comparable to that of a deputy chief physician [6][7]. - The company has integrated various data types (voice, images) into its model, enhancing its ability to provide comprehensive patient services [8]. - iFlytek Medical's big model has been trained on a vast dataset, creating a **data flywheel** effect that continuously improves its performance through real-world applications [19][21]. Market Position and Strategy - iFlytek Medical aims to leverage its AI capabilities to enhance healthcare delivery across G-end (government), B-end (hospitals), and C-end (patients) sectors, establishing a diversified product matrix [9][10]. - The company has established partnerships with top hospitals, enhancing its credibility and market reach. It has covered **31 provinces** and provided services to over **910 million** patients [10][11]. - The government has shown support for AI in healthcare, with policies promoting the application of big models in various sectors, which bodes well for iFlytek Medical's growth [10]. Challenges and Opportunities - The company faces competition from other AI healthcare providers but believes its proprietary technology and data capabilities give it a competitive edge [31][34]. - iFlytek Medical is focused on expanding its services in both urban and rural healthcare settings, addressing the need for balanced medical resource distribution [10][36]. - The company is also exploring opportunities in post-discharge patient management, which is becoming increasingly important as healthcare systems emphasize continuity of care [53][54]. Other Important Content - iFlytek Medical's AI solutions have been implemented in various hospitals, including **Qilu Hospital** and **Shandong Provincial Hospital**, demonstrating the practical application of its technology [15][16]. - The company has developed a comprehensive patient management system that includes follow-up care and personalized health plans, which has shown to reduce readmission rates significantly [55][56]. - iFlytek Medical's collaboration with **Huawei** focuses on providing computational power and data centers, enhancing its operational capabilities in the healthcare sector [43][44]. This summary encapsulates the key insights from the conference call, highlighting iFlytek Medical's financial performance, technological advancements, market strategies, and the challenges and opportunities it faces in the evolving healthcare landscape.