投资者提问： 1.公司的模型现在还是基于全国产'底座训练吧？ 2.讯飞星火有拓展文生视频的能力的规划么？ 免责声明：本信息由新浪财经从公开信息中摘录，不构成任何投资建议；新浪财经不保证数据的准确 性，内容仅供参考。 董秘回答(科大讯飞SZ002230)： 您好，公司自开展讯飞星火大模型攻关以来，始终以对标国际最先进大模型为目标，坚持走自主可控发 展路线，在算力较为紧张的情况下保持了全栈自主可控的星火底座大模型持续对标全球第一阵营，并且 在教育、医疗、消费者等重点赛道不断取得商业化落地成果。未来，公司将围绕算法、数据及应用场景 需求等持续提升星火大模型的多模态能力，并根据研发进度及产品规划，适时推动技术与应用的融合落 地。感谢您的关注。查看更多董秘问答>> ...

1月15日，科大讯飞AI营销（AIMarX）在深圳正式发布基于SuperAgent构建的营销智能体平台，主要解 决需要智能规划与多方协同的复杂营销场景。 ...

本报（chinatimes.net.cn）记者石飞月 北京报道 企业服务市场正迎来AI深度重构的关键时刻。1月13日，科大讯飞推出"招采智能体平台"，并将其定位 为垂直领域的"智能体工厂"。自进入AI 2.0时代以来，科大讯飞就基于大模型在B端市场"高歌猛进"，其 相关项目的中标数量和金额在行业中均排名第一，并带动营收突破纪录。不过，技术一路狂奔的另一 面，业内人士指出，数据安全之虑、场景适配之难、商业模式之变，仍构成一道道待解的考题。 瞄准招标采购赛道 既然取名为"招采智能体平台"，其核心技术自然是智能体。科大讯飞企业数字化业务群CTO张永亮介 绍，通过"招标文件智能编审""异常行为检测""辅助清标评标"等一系列场景化智能体，平台能够将招标 文件编制时间从平均5-7个工作日压缩至30分钟，围串标检测准确率达到96%，整体评标效率提升70%以 上。 作为企业运营的核心环节，当前招标采购面临着效率提升难、合规管控难、成本优化难的多重挑战，而 随着数字化转型向深水区迈进，招采环节的智能化升级已成为企业的普遍需求。数据显示，我国企业采 购市场规模庞大，数字化采购的渗透率仍有较大提升空间。招采环节的智能化不仅能直接降 ...

Investment Rating - The report assigns an "Overweight" rating for the industry [4]. Core Insights - The transition from traditional SEO to GEO (Generative Engine Optimization) represents a paradigm shift in how brands are perceived and trusted in AI-driven search environments. GEO focuses on enhancing the credibility and citation frequency of brands in AI-generated answers, moving beyond mere visibility to being actively referenced by AI [2][8]. - The market potential for GEO is projected to reach a "billion-dollar level," driven by the replacement of existing SEO budgets and new allocations for AI search [2][21]. Summary by Sections 1. Definition and Essence of GEO - GEO is defined as an optimization strategy that ensures brands and content are actively mentioned in AI-generated answers, contrasting with traditional SEO which focuses on ranking [8]. - The emergence of GEO is attributed to the rise of AI search, which bypasses traditional click-through processes, leading to a significant drop in natural click rates [10][11]. 2. Technical Principles: Trust Engineering on the RAG Link - The RAG (Retrieval-Augmented Generation) architecture is central to GEO, shifting the focus from keyword matching to semantic understanding and trust-building [14][15]. - GEO aims to enhance content visibility, retrievability, and trustworthiness, rather than simply improving rankings [15]. 3. Market Space: SEO Replacement and New AI Search Demand - The global SEO service market is estimated at approximately $80 billion in 2024, with GEO expected to capture 10-20% of this budget, alongside new AI search allocations, leading to a potential market size exceeding $100 billion by 2030 [16][19]. - In China, the GEO market is projected to grow from 2.9 billion yuan in 2025 to 24 billion yuan by 2030, reflecting a CAGR of about 52.4% [21]. 4. Business Model: Transition from Labor-Intensive Services to Technology Platforms - The current GEO service model is primarily project-based, but it is expected to evolve towards a subscription-based SaaS model combined with performance-based pricing [22][25]. - The anticipated gross margin for GEO services is expected to rise significantly, aligning with the characteristics of the high-tech, high-concentration software industry [25]. 5. Investment Recommendations: Mapping the Content-Knowledge-Retrieval-Computing Chain - The report identifies key investment targets across the GEO value chain, including companies like Mifus, Minglue Technology, and iFlytek, which are positioned to benefit from the shift towards AI-driven marketing and content creation [27][30].

Core Insights - The "2025 Technology Wind List" annual event was held on January 15, 2026, in Beijing, with the theme "Inspiring New Intelligence, Embarking on a New Journey" [1][4] - Key participants in the roundtable discussion on "New Paradigm of Agent Collaboration" included CTOs from various companies, highlighting the importance of AI in different sectors [1][4] Group 1: AI Application Insights - Zhang Yongliang, CTO of iFLYTEK, emphasized the significance of identifying the right scenarios for AI application, particularly in sectors like education, finance, and healthcare, where efficiency can be improved or repetitive tasks can be automated [1][4] - The effectiveness of AI tools is paramount, and it should be validated through data to demonstrate their impact [5] Group 2: Integration and Cost Efficiency - A critical concern for enterprises is avoiding the complete overhaul of existing systems; instead, the focus should be on seamless integration of AI technologies with current systems to enhance efficiency and reduce development costs [5]

| 代码 | 简称 | 今日涨跌幅（%） | 今日换手率（%） | 主力资金流量（万元） | | --- | --- | --- | --- | --- | | 600410 | 华胜天成 | 10.00 | 3.27 | 59365.36 | | 002410 | 广联达 | 5.58 | 13.45 | 46718.12 | | 600271 | 航天信息 | -0.51 | 5.29 | 10222.13 | | 688631 | 莱斯信息 | 1.53 | 13.48 | 9937.29 | | 688111 | 金山办公 | -2.10 | 2.83 | 8703.31 | | 300479 | 神思电子 | 12.05 | 29.11 | 7091.08 | | 300448 | 浩云科技 | 3.52 | 11.66 | 7056.62 | | 002230 | 科大讯飞 | 1.20 | 6.85 | 6472.65 | | 600536 | 中国软件 | -0.22 | 3.74 | 6201.35 | | 300079 | 数码视讯 | 0.33 | 7.31 | 5762.21 | | ...

TCL创始人、董事长李东生聚焦科技赋能实体经济的核心逻辑。他指出，当前全球科技竞争日趋激烈， 人工智能正从概念走向深度产业化，成为未来经济增长的关键力量。TCL始终以技术创新为核心驱动 力，持续加大AI研发投入，其发布的星智大模型3.0已在半导体显示行业充分应用，显著提升产品开发 与材料研发效率。未来，TCL将进一步推动AI技术渗透生产研发核心环节，打造行业领先的AI智能体中 心，为制造业高质量发展提供坚实技术支撑。作为深商群体的重要代表，李东生还积极参与深商服务网 络建设，为区域产业协同发展助力。 专题：第九届深商盛典暨中国企业家俱乐部20年活动 1月12日，深圳前海国际会议中心灯光璀璨、智慧云集，第九届深商盛典暨中国企业家俱乐部20年活动 在此隆重举行，一场汇聚中国顶尖企业家的思想盛宴如期绽放。活动由复旦大学国际关系与公共事务学 院副教授、知名主持人蒋昌建全程主持，李东生、林锋、谢永林、何小鹏、刘庆峰、林立、马蔚华、张 文中、尹烨等一众行业领军者齐聚鹏城，以"遇见未来的中国企业思想"为核心主题，立足国家高质量发 展战略全局，围绕产业革新、技术突破、生态构建、社会责任等关键议题深入探讨。在全球技术浪潮加 速 ...

科大讯飞方面表示，未来，第三方开发者、行业专家可以基于此平台，开发并上架更丰富的垂直场景智 能体，共同打造招采领域的"智能应用商店"。 （文章来源：新京报） 近日，4名评标专家因为送餐延误导致身体不适的消息引发关注。官方通报此事时表示"评标工作责任 重、强度高，评标时长不确定"。值此当口，科大讯飞宣布推出"招采智能体平台"，在降低合规风险的 同时提升评标工作效率。 1月15日，科大讯飞方面告诉贝壳财经记者，招采业务场景千变万化，因此平台的终极目标是构建生 态，其内置了覆盖招标、投标、评标核心流程的数十个专业Agent。 在广受关注的评标领域，科大讯飞工作人员向记者展示了通过AI自动评审与推荐的流程。记者观察 到，AI可以自动对标书进行文本解析，数据提取和图像识别，并在文件评审流程进行多维度对比和AI 评分，最终可以输出一份智能评标报告，以及供应商优劣对比表和推荐的中标候选人（附理由）。 ...

Group 1 - The software leading ETF (159899) has seen a net inflow of funds for three consecutive trading days, accumulating a total scale of 290 million yuan [1] - This ETF closely tracks the CSI Software Index, covering leading companies in various segments such as basic software, application software, and industrial software, with AI application-related concepts accounting for 35% of its weight [1][2] - The top ten holdings in the index include companies like Tuowei Information and Kingsoft Office, which are expected to benefit from the industry boom driven by AI technology [1] Group 2 - The AI application sector is experiencing a new wave of technological breakthroughs and ecosystem construction, highlighted by the launch of new features in the Qianwen APP, which has surpassed 10 million monthly active users within two months of its launch [2] - Ant Group has also introduced a new product called "Wuli," further expanding the AI application ecosystem [2] - Policy support is evident, with recent publications emphasizing the importance of application-driven development in the AI industry, and successful AI implementations in regulatory systems showcasing its potential in governance [3] Group 3 - Significant progress in technological self-innovation is marked by the collaboration between Zhipu and Huawei to open-source the first multi-modal SOTA model trained on domestic chips, indicating a breakthrough in AI infrastructure [3] - Market analysis suggests a shift in algorithm mechanisms from pursuing "traffic" to building "trust," which may lead to a systematic re-evaluation of content value from official media and professional publishing companies [3] - As domestic large models penetrate multiple scenarios, the AI application sector is entering a phase of accelerated industrialization [3]

1月14日消息，科大讯飞在北京正式发布"招采智能体平台"，将其定位为垂直领域的"智能体工厂"。平 台基于讯飞自研的"星辰Agent"技术底座，企业可以通过低代码或零代码方式组装AI能力组件，构建贴 合自身业务流程的专属智能体。 ...