1、外媒爆料：美军正做准备，一旦特朗普下令攻击伊朗，行动可能持续数周 路透社当地时间13日独家报道称，两名美国官员表示，美国军方正就美国总统特朗普一旦下令攻击 伊朗后，可能持续数周的行动做准备。这可能演变为两国间前所未有的严重冲突。路透社称，美官员披 露的这一信息使美伊正在进行的外交努力面临更高风险。路透社称，上述官员表示，这一次正在进行的 行动规划更为复杂。其中一名官员表示，在持续的军事行动中，美军不仅可能打击伊朗的核基础设施， 还可能打击其国家机构和安全设施，该官员拒绝透露具体细节。这名官员还表示，美国方面完全预计伊 朗会进行报复，从而在一段时间内形成来回打击与报复的循环。（环球网） 2、金饰价涨到1565元 当地时间13日，美国CPI数据公布后，市场对美联储降息的概率预期上升，美元指数回落，支撑贵 金属资产吸引力。现货黄金涨2.41%，报5042.81美元/盎司；COMEX黄金期货涨2.33%，报5063.8美元/ 盎司。现货白银涨2.81%，报77.34美元/盎司；COMEX白银期货涨2.10%报77.27美元/盎司。14日，国内 品牌金饰价格随之上调，周生生足金饰品报1551元/克，较前一日1524 ...

责任编辑：钟离 2月14日，武汉市人民政府印发《武汉市支持人工智能OPC创新发展若干措施》。《若干措施》提出， 提供创业启动支持。支持人工智能领域人才在汉创办OPC，综合评定其经营发展、技术创新情况，给予 初创企业最高100万元的创业资助。OPC生态社区构建与企业质量和发展阶段相匹配的支持政策体系， 对入驻项目从技术创新性、成熟度、市场潜力等维度开展综合评定，梯度匹配涵盖租金减免、启动资 金、"拨转股"、长期租金转股权等形式的支持政策。各区整合区内中试平台、概念验证中心及各类中介 服务资源，为符合条件的OPC提供最长2年的免费公共服务包，涵盖财务代账、税务申报、法律咨询及 专利申报等。 股票频道更多独家策划、专家专栏，免费查阅>> ...

责任编辑：钟离 股票频道更多独家策划、专家专栏，免费查阅>> 2月14日，武汉市人民政府印发《武汉市支持人工智能OPC创新发展若干措施》。《若干措施》提出， 支持模型工具研发应用。鼓励各类创新主体研发轻量化开发工具、低代码平台及垂直领域模型工具，攻 关智能体高效开发框架；针对OPC在垂直行业落地中面临的技术瓶颈，支持OPC出题，高校院所、新型 研发机构揭榜答题，实现技术供给与产业需求精准对接与协同研发，对经评定符合条件的项目，纳入市 级重点研发计划，给予最高100万元资金支持。支持OPC打造垂直领域模型，对于性能先进的模型，依 规对研发单位按照研发成本的30％给予最高500万元补助。支持OPC通过调用第三方大模型API（应用 程序编程接口）或者采用本地化部署等方式，开展大模型垂直领域应用，依规按照不超过模型采购费用 的50％给予最高100万元补助。 ...

过去一年，在AI核心业务推动下，科大讯飞出海提速，交出了一份稳健增长的成绩单。2026年2月13 日，在公司年会上，科大讯飞董事长刘庆峰、总裁吴晓如分别发表演讲，系统总结2025年经营成果，并 部署2026年战略重点。 2025年，科大讯飞在资源受限的情况下，不仅实现了归母净利润和经营性现金流的扎实增长，更通过攻 克国产卡训练难关，推动星火大模型持续迭代，近日发布的星火X2大模型在数学、推理、医疗等关键 能力上对标国际一流模型水平。此外，公司去年大模型相关项目中标额超23亿元，再度蝉联大模型"标 王"。 根据科大讯飞此前发布的2025年度业绩预告，公司预计2025年归属于上市公司股东的净利润为7.85亿元 至9.5亿元，较上年同期增长40%至70%；扣除非经常性损益后的净利润为2.45亿元至3.01亿元，同比增 长30%至60%。 在回顾2025年业绩时，吴晓如在年会上披露，2025年科大讯飞经营回款超270亿元，经营性现金流再创 新高达30亿元、同比增长28%。产品端同样亮眼：开放平台大模型日均调用量较上年猛增45倍，毛利增 长244%，单业务毛利首次突破10亿元；讯飞听见用户规模突破1亿，AI营销海外毛利 ...

Core Viewpoint - The recent release of Seedance 2.0 by ByteDance has significantly narrowed the gap between its AI models and those of leading companies like Google, with the difference now estimated to be as little as one to two months [62][60]. Group 1: Seedance 2.0 - Seedance 2.0 has generated excitement in the AI community, with many users expressing shock and admiration for its capabilities [64][66]. - The model demonstrates strong instruction-following abilities, effectively understanding complex prompts and generating high-quality video content [71][72]. - Users report that Seedance 2.0 has surpassed previous models in terms of performance, making it suitable for creating professional-quality animations and videos [73][74]. Group 2: Seedream 5.0 Lite - Seedream 5.0 Lite, the latest image model from ByteDance, has improved in two key areas: subject consistency and instruction-following ability [20][78]. - Users have noted that the model generates images with better consistency, reducing the "out-of-place" feeling previously experienced with earlier versions [21][78]. - The model's ability to follow complex instructions has been highlighted as a significant advancement, making it easier for users to edit images effectively [82]. Group 3: Doubao Model 2.0 - Doubao Model 2.0 has shown significant improvements in complex reasoning and agent tasks, outperforming its predecessor by a considerable margin [26][83]. - The model is designed to handle multi-modal tasks natively, integrating text, images, and video without the need for separate plugins, which enhances its efficiency [31][87]. - Doubao 2.0 has also reduced inference costs significantly, making it more accessible for commercial applications, particularly in agent scenarios where token consumption is high [45][99]. Group 4: Strategic Positioning - ByteDance's approach to AI development closely resembles that of Google, focusing on integrating models with applications to create a feedback loop that informs future model improvements [100][104]. - The company leverages its large user base and content creators to identify gaps in capabilities, allowing for targeted enhancements in its AI models [102][103]. - The synergy between model development and cloud services, particularly through Volcano Engine, positions ByteDance favorably in the competitive landscape [108][109].

Investment Rating - The report maintains a strong buy rating for the media industry, indicating a positive outlook for investment opportunities in the AI video sector [2]. Core Insights - The AI video generation technology is evolving rapidly, transitioning from GAN to DiT architectures, which are crucial for advancing towards AGI. This evolution is expected to significantly enhance the capabilities of AIGC (AI-Generated Content) [3][9]. - The global AI video generation market is projected to reach $296 million by 2026, with a year-on-year growth of 35.16%. The industry is exploring both consumer (C-end) and business (B-end) revenue models, with significant advancements in commercial applications expected in the near future [3][4]. Summary by Sections 1. Video Generation Evolution - Video generation integrates multiple modalities, including text, images, and audio, which enhances its complexity and expressiveness, representing the upper limit of AIGC capabilities [7]. - The technology has progressed from early GAN models to the current DiT architecture, marking a significant turning point in the industry with the introduction of models like OpenAI's Sora [9][25]. 2. Technical Progress - Current AI video generation models can produce short segments that approach professional production quality, with resolutions supporting 1080p and frame rates reaching 30fps. However, challenges remain in generating longer videos and maintaining physical realism [34][36]. - The emergence of world models is anticipated to address existing limitations in video generation, potentially leading to a new phase of technological advancement [33]. 3. Commercialization Progress - The AI video generation market is expanding rapidly, with both consumer and business segments progressing simultaneously. The C-end focuses on subscription models, while the B-end primarily utilizes APIs for applications in advertising and e-commerce [3][4]. - The industry is witnessing a shift towards integrating AI capabilities into film production, with significant projects already generating substantial revenue, such as Utopai's projects totaling approximately $110 million [3][4]. 4. Core Beneficiaries - Key companies benefiting from this trend include technology firms with proprietary algorithms, content providers with extensive asset libraries, and platforms actively integrating AI into marketing strategies [4].

Core Insights - Ant Group has developed and open-sourced the world's first trillion-parameter thinking model, Ring-2.5-1T, which achieves fast inference speed, deep reasoning capabilities, and excellent long-range task execution [1][9] - The model scored 35 out of 42 in the IMO competition and 105 in the CMO, significantly exceeding the national training team's score line [1][7] Model Architecture - Ring-2.5-1T is based on the Ling 2.5 architecture, utilizing a hybrid linear attention mechanism that combines MLA (Multi-Head Latent Attention) and Lightning Linear Attention in a 1:7 ratio [2] - The model's active parameter count increased from 51 billion to 63 billion, yet its inference efficiency improved due to its linear time complexity [2] Performance and Capabilities - The model demonstrates significant advantages in long-sequence reasoning tasks compared to other models with similar parameter counts, particularly in throughput as sequence length increases [2] - Ring-2.5-1T has been benchmarked against various models and has achieved optimal performance in high-difficulty reasoning tasks and long-duration task execution benchmarks [5] Training Innovations - The model incorporates a dense reward mechanism based on Reinforcement Learning with Verifiable Rewards (RLVR), enhancing its logical reasoning and proof techniques [4] - It also employs large-scale fully asynchronous Agentic RL training, improving its autonomous execution capabilities in complex tasks [4] Ecosystem and Future Developments - Ring-2.5-1T is compatible with major intelligent agent frameworks and has been made available on platforms like Hugging Face and ModelScope [7] - Ant Group has also released other models, including LLaDA2.1 and Ming-flash-omni-2.0, focusing on various AI capabilities such as non-autoregressive parallel decoding and multimodal representation [8] - The company aims to provide reusable foundational solutions for developers, with plans to expand into video understanding, complex image editing, and real-time audio generation [8]

Core Insights - ByteDance officially launched the Doubao-Seed-2.0 series on February 14, focusing on optimizing large-scale production environments and enhancing the execution capabilities for complex real-world tasks [1] Group 1: Model Performance - The flagship Doubao-2.0 Pro model demonstrated top-tier foundational capabilities, winning gold medals in IMO, CMO mathematics competitions, and the ICPC programming competition, while outperforming Gemini 3 Pro in the Putnam benchmark test [1] - In terms of knowledge coverage, Doubao-2.0 has strengthened its long-tail domain knowledge, achieving comparable scores in scientific knowledge tests with Gemini 3 Pro and GPT 5.2, and ranking among the industry leaders in interdisciplinary knowledge application [1] Group 2: Multimodal Capabilities - Doubao-2.0 has undergone a comprehensive upgrade in multimodal capabilities, achieving the best industry scores in visual reasoning, spatial perception, and long-context understanding tests, enabling efficient comprehension of charts, complex documents, and videos [1] - The model has enhanced its understanding of time series and motion perception, allowing for real-time video stream analysis, environmental awareness, and proactive interaction, applicable in diverse life scenarios such as fitness guidance and companionship [1] Group 3: Agent Capabilities and Pricing - The Agent capabilities of Doubao-2.0 Pro excelled in instruction adherence and tool invocation assessments, achieving the highest score of 54.2 in the HLE-Text evaluation, surpassing other models [2] - Doubao-2.0 Pro is now available on the Doubao App, desktop client, and web version, with a pricing structure of 3.2 yuan per million tokens for input (up to 32k) and 16 yuan per million tokens for output, offering better cost-effectiveness compared to Gemini 3 Pro and GPT 5.2 [2] - The Lite version is priced at only 0.6 yuan per million tokens, with overall performance exceeding that of the previously released Doubao-1.8 [2]

临近春节，字节跳动AI好事频发。 2月14日，豆包大模型2.0发布，这是字节跳动最新推出的多模态Agent模型，也是豆包大模型自2024年5 月正式发布以来首次大版本的跨代升级。 与此同时，豆包新一代AI视频生成模型Seedance2.0及图像创作模型Seedream5.0 Lite的好评如潮，标志 着字节跳动已实现多模态Agent、AI图像、视频创作三大维度的全栈式技术突破，全链路AI布局优势日 益凸显。 豆包2.0推理能力登顶，成本优势凸显 此次升级，豆包2.0系列包含Pro、Lite、Mini三款通用Agent模型和Code模型。全面适配从深度推理到成 本敏感的不同业务场景。 整体看，豆包2.0可以进行更强多模态理解：在多模态感知、高精度文字提取、图表理解、空间理解、 运动理解、视觉知识和推理、长视频理解等方面表现出色。 值得注意的是，豆包2.0还进一步降低了推理成本，这也一直是豆包的优势之一。其模型效果与业界顶 尖大模型相当，但token定价降低了约一个数量级。在目前的AI大战中，这一成本优势将变得更为关 键。 AI技术领跑，视频与图像全球影响力破圈 随着三大模型的集中升级，字节跳动的AI全链路布局优 ...

Group 1 - The core point of the news is that Qunhe Technology, one of the "Six Little Dragons of Hangzhou," has made significant progress towards its IPO in Hong Kong, having received the overseas issuance listing notice from the China Securities Regulatory Commission [1] - Qunhe Technology is expected to officially list in 2026, becoming the first company among the "Six Little Dragons of Hangzhou" to go public, following a year since its IPO application was submitted to the Hong Kong Stock Exchange in February 2025 [1] - The company owns the spatial design platform "CoolJia" and the next-generation spatial intelligence solution "SpatialVerse," which focuses on AI development for indoor environments and virtual product displays [1] Group 2 - In addition to Qunhe Technology, other companies among the "Six Little Dragons of Hangzhou" are also actively preparing for their IPOs, indicating a broader trend of capital market engagement within this group [1][4] - Yunshe Technology has submitted IPO counseling materials to the Zhejiang Securities Regulatory Bureau, with the counseling period expected to conclude between April and June 2026, potentially leading to an A-share application [2] - Strong Brain Technology, a brain-computer interface company, is rumored to have secretly submitted an IPO application to the Hong Kong Stock Exchange [3]