Core Viewpoint - The company has completed a full industry chain layout for "edge perception - digital middle platform - industry large model - industry application robot" and is accelerating the research and development of machine vision large models and industry-specific large models [1] Group 1: Industry Chain Layout - The company has launched the Zhongke Jiyun government affairs large model aimed at government agencies [1] - A multi-modal industrial inspection large model has been developed for industrial inspection scenarios [1] - The company has independently developed a medical large model to support the research and iteration of intelligent anesthesia robots [1] Group 2: Technological Applications - The models empower medical imaging analysis and tumor target delineation in diagnostic assistance [1] - The company plans to further strengthen internal common technology focus and technological collaboration [1] - The aim is to cover core business areas such as smart government affairs, intelligent manufacturing, and smart healthcare, providing technical support for product upgrades and digital transformation across various industries [1]

证券日报网1月13日讯，新开普（300248）在接受调研者提问时表示，未来公司在大模型技术领域将聚 焦三大核心布局方向，持续深化技术应用与价值转化：1、深耕校园数据洞察与智能问答场景：持续迭 代优化面向校园场景的自然语言处理(NLP)搜索模型，重点提升"校园问数"系统的意图识别、自然语言 生成及权限计算和隐私计算能力，实现校园数据实时查询、智能追因与洞察预测的高精度响应，为高校 信息化管理决策提供高效支撑。2、发力视觉模型研发与产教融合场景落地：重点推进视觉模型、视频 物体检测模型在产业学院教学、实训场景的应用，同步攻坚模型轻量化技术，将相关模型部署至边缘计 算设备，实现本地化、低延迟的AI辅助教学，助力产教融合模式创新升级。同时，推进AI+教学融合实 践，开展AI出题、智能批阅、个性化学习推荐、AI数字人视频生成等多元应用探索，并布局标准化AI 教学产品对外输出，构建从B端到C端的价值闭环，进一步拓宽盈利边界，提升长期发展韧性。3、强化 技术能力复用与跨场景商业化延伸：将在校园场景中验证成熟的AI技术能力，横向复制拓展至智慧园 区、智慧农业、智慧水务的"物联网+AI、门户+AI、数据治理+AI、终端+AI" ...

Core Insights - The rise of AI technology has made digital human short videos a new trend in content creation, with creators and companies seeking to reduce costs and improve efficiency, but facing challenges in core preparation, visual quality, and tool selection [1] Group 1: Preparation and Foundation - Proper preparation is crucial for the final output, as outlined by the national broadcasting standards which require digital humans to have appropriate appearances and smooth movements without audio-video delays or lip-sync issues [3] - Key production steps include accurately replicating the human image and voice, ensuring natural sound quality that matches the digital human's character, and avoiding distortions or delays [3] - Coordination between actions and scenes is essential, with digital human movements needing to align with content logic and the ability to switch between real and virtual backgrounds to enhance immersion [3] Group 2: Content and Material Organization - Clear content positioning and requirements must be established, determining the digital human's style based on usage scenarios, and ensuring that the script is concise and fits within a 300-word limit [4] - Preparation of materials includes obtaining clear frontal images and approximately one minute of real human audio for accurate replication, along with relevant background materials for scene-based content [4] Group 3: Tool Selection and Efficiency - The choice of tools significantly impacts production efficiency and quality, with a recommendation to select user-friendly tools that meet specific needs [6] - Among the leading platforms, Shengnuo Kexinda stands out for its comprehensive advantages, and mastering core techniques can help avoid mistakes and reduce post-production costs [6] Group 4: Post-Production and Distribution - Post-production optimization and distribution, while not core production steps, can enhance the reach and value of digital human short videos by refining content and adapting it to platform requirements [8] - Shengnuo Kexinda's tools facilitate efficient post-production and distribution, allowing for batch generation of videos without real human appearances, thus saving time and labor costs [8] Group 5: Tool Comparisons - Shengnuo Kexinda offers a complete workflow for short video production, featuring capabilities like voice cloning and AI face-swapping, supporting high-fidelity 4K outputs and multi-action displays [9] - Baidu Smart Cloud Xiling provides high-precision facial restoration but has limitations in cross-platform compatibility and is more suited for large enterprises [9] - Tencent Zhiying integrates deeply with the WeChat ecosystem, offering user-friendly templates but lacking in batch production capabilities [9] - Jianying Digital Human is a widely used editing tool with strong usability but limited core functionalities and high-fidelity effects [9] Group 6: Industry Outlook - The core of digital human short video production lies in balancing quality and efficiency while adhering to industry standards, with the right tools addressing challenges like the need for real human appearances and batch production [11] - New creators and corporate teams can efficiently produce high-quality digital human short videos by mastering the outlined methods and selecting appropriate tools, as technology continues to mature and expand application scenarios [11]

Core Viewpoint - The A-share market experienced a rapid decline in AI application concept stocks, indicating a potential shift in investor sentiment towards these companies [1] Group 1: Company Performance - Hengxuan Technology saw a decline of over 9% [1] - Guanghetong dropped by more than 8% [1] - Lexin Technology fell by over 6% [1] - Mango Super Media, Tuo Wei Information, Ruixin Micro, CloudWalk Technology, Hongsoft Technology, and Runhe Software all experienced declines of over 5% [1]

Core Insights - Banqup Group SA is in advanced negotiations to sell all shares in its operational Baltic companies to Fitek Oü, with a target completion date by the end of February 2026, pending approval from competition authorities in Estonia, Latvia, and Lithuania [1][5]. Group 1: Transaction Details - The proposed transaction involves an enterprise value of €9.5 million, subject to adjustments based on the net financial cash/debt position as of December 31, 2025 [5]. - The transaction is expected to enhance Banqup's focus on core digital services and optimize its print operations [3][5]. - A partnership agreement is also being negotiated, allowing Fitek to become an authorized reseller of Banqup's digital product suite in the Baltic States [5]. Group 2: Financial Performance - The Baltic companies generate an annual EBITDA of €1.3 million, with a product suite that includes €0.3 million in Digital Banqup revenue, €4.0 million in digital legacy revenue, and €10.0 million from print-related services [5]. - The transaction proceeds are anticipated to strengthen Banqup's balance sheet and working capital position [5]. Group 3: Strategic Focus - The divestment reflects Banqup's strategic focus on growing its core digital services in key European markets [3]. - The acquisition by Fitek is expected to enhance capabilities in digital business solutions and improve service offerings in the Baltic region [3].

Financial Performance - The group's unaudited consolidated revenue decreased in Q4 2025 by 2% year-over-year to €3,525 thousand, compared to €3,598 thousand in Q4 2024 [1] - For the full year 2025, total unaudited consolidated revenues decreased by 7% year-over-year to €13,549 thousand, down from €14,533 thousand in 2024 [2] - If exchange rates had remained the same as in 2024, Q4 2025 revenue would have been €3,786 thousand, reflecting a 5% increase, while total revenue for 2025 would have been €14,015 thousand, indicating a 4% decrease [1][2] User Metrics - In Q4 2025, the volume of SMS messages sent increased by 14% year-over-year, reaching 70,665 thousand pieces, while the number of active users decreased by 5% to 18,876 [3][4] - For the full year 2025, the volume of SMS messages sent was 261,931 thousand pieces, a 5% increase from 249,228 thousand in 2024, but the number of active users declined by 9% to 23,922 [3][4] Average Revenue Per User (ARPU) - The average revenue per user (ARPU) for Q4 2025 was €187, a 3% increase from €182 in Q4 2024 [3] - For the full year 2025, ARPU was €566, up 3% from €549 in 2024 [3] Market Developments - The decline in active users was primarily due to regulatory changes in the United States that restricted unregistered campaigns, while the UK saw an increase in active users during 2025 [5] - Growth in revenue was driven by developments in the UK and Australian markets [4] New Features and Initiatives - In Q4 2025, the Textmagic platform introduced an automation feature for creating automated messaging systems, enhancing customer interaction efficiency [6] - Touchpoint.com was launched as a new platform for creating professional email designs using AI, aimed at broadening the customer base and supporting Textmagic users [7] Operational Measures - To improve cash flow and profitability, operational measures were implemented in Q4 2025, including optimizing development costs and discontinuing operations of the Romanian subsidiary [8]

Group 1 - The A-share market opened with all three major indices rising, with the Shanghai Composite Index up by 0.11% to 4169.7 points [1] - The Shenzhen Component Index increased by 0.21%, while the ChiNext Index saw a rise of 0.07% [1] - AI application concept stocks continued their upward trend, with companies like Yidian Tianxia, Tongda Hai, and Di'an Diagnosis (300244) hitting the 20% daily limit [1] Group 2 - The military equipment sector experienced a correction [1]

21智讯1月13日电，新开普在投资者关系活动中表示，未来两到三年公司将在大模型技术领域聚焦三大 方向：一是持续优化校园场景NLP搜索模型，提升'校园问数'系统的意图识别、生成能力和隐私计算水 平，支持数据智能查询与决策分析；二是推进视觉模型与视频物体检测模型在产业学院实训中的应用， 攻坚轻量化技术并部署至边缘设备，实现低延迟AI辅助教学；三是将校园验证成熟的'物联网+AI、门 户+AI、数据治理+AI、终端+AI'能力复制至智慧园区、智慧农业、智慧水务等B端行业，构建可复用的 技术中台，推动跨场景商业化落地。 ...

Group 1: Company Overview and Market Position - Newcap Electronic Co., Ltd. is a leading company in the information technology sector for higher education in China, focusing on smart campus solutions and educational management information systems [2] - The company has served over 1,400 universities, achieving a market share exceeding 40% in the smart campus sector [2] Group 2: AI Strategy and Development - In 2025, the company's core development strategy will center around AI, with its self-developed large language model, "Xingpu Model," already implemented in over 30 universities [3] - The AI solutions include applications in data governance, intelligent Q&A, and campus assistants, which have received multiple provincial innovation application awards [3] Group 3: Revenue Model and Future Growth - The current revenue model for AI services is based on traditional project bidding, ensuring quality and efficiency in service delivery [4] - The company plans to establish a dual revenue growth system, leveraging opportunities from the "14th Five-Year Plan" for information technology upgrades in universities [4] Group 4: Technical Focus Areas - Future technical focus will include enhancing natural language processing for campus data insights and intelligent Q&A systems [4] - The company aims to develop visual models for educational applications and explore AI integration in teaching practices, including personalized learning recommendations [4] - There will be an emphasis on reusing technical capabilities across different sectors, such as smart agriculture and smart water management, to broaden the company's business growth boundaries [4]

摘要 Q&A 新开普在 2025 年的整体业务情况如何？ 2025 年，新开普在高效线下信息化建设方面取得了显著进展。财政和银行资 金投入较 2023 年和 2024 年有明显提升，订单增长显著。然而，由于 To B 业 务的实施周期较长（6 至 12 个月），对 2025 年整体业绩的正向影响有限，但 预计将对 2026 年的各期报告产生积极影响。此外，公司在激烈竞争中新增了 30 多个公办本科院校客户，这些优质客户资源为公司带来了稳定的收入来源。 新开普在人工智能领域有哪些重要进展？ 2025 年，新开普围绕人工智能技术进行了多项尝试。公司首个本地化部署 AI 客户湖南科技大学之后，累计服务了 30 多所高校，包括西安交通大学、山西 大学、上海电影学院和福州大学等。这些高校采用了新开普的 AI 数据指标、数 据治理、门户及 Agent 服务，并获得了良好效果。此外，公司还通过产教融合 新开普 20260112 新开普 AI 业务增长显著，2025 年湖南科技大学项目合同额从百万级别 增至千万元级别，西安交通大学等多个 AI 应用项目也呈现增长态势，预 计 2026 年将实现指数级增长。 新开普通过与高校签 ...