[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fXZNCheQNlxyXML7nCAsd94F2XknqqXt55it9O6XJ460":3},{"code":4,"msg":5,"data":6},200,"操作成功",{"id":7,"title":8,"content":9,"digest":10,"source":10,"coverPath":11,"thumbsCoverPath":12,"isTop":13,"isShow":14,"baseClick":13,"clickCount":15,"createTime":16,"typeId":17,"isNewest":18,"newsInfoTypeRespVo":19,"voiceUrl":22,"voiceSize":23,"taskId":24,"releaseTime":25,"titleEn":26,"contentEn":27,"voiceUrlEn":28,"taskIdEn":29,"voiceSizeEn":30},1402,"2025人工智能计算大会观察：token成本成AI应用规模化瓶颈 行业寻求算力“破局”","\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">《关于深入实施“人工智能+”行动的意见》印发至今刚满一月，产业界的“起跑”已然加速。\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">昨日在北京举行的2025人工智能计算大会，如同一个绝佳的观察窗口。财联社记者在现场注意到，与会专家与企业代表的讨论焦点，已从宏观的政策解读转向了具体的“施工蓝图”。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F09\u002F2b0181d29e774b09a170933f336d48f2\u002FAA1NoDZp.jpg\" width=\"556\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">本届大会紧扣人工智能基础设施建设和国产AI算力体系优化，聚焦推动算法创新与应用落地，以算力核心要素为创新牵引，汇聚产学研用各界力量，共同推动人工智能产业高质量发展。现场，中国移动、浪潮信息、智源研究院、昆仑芯等30多家企业和机构，联合发布《基于超节点创新联合体，打造行业智能体——智算应用“北京方案”》，率先响应国家《关于深入实施“人工智能+”行动的意见》。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong style=\"font-size: 18px; color: rgb(255, 153, 0);\" class=\"ql-lineHeight-1-75\">“实现跨地域、跨硬件的算力连接与普惠共享”\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">会上，西湖大学“可信及通用人工智能实验室”创始人、欧洲科学院院士金耀初梳理了人工智能发展的主要脉络，并指出其发展路径与人脑智能的涌现过程相似——经历了演化、发育与学习的三个关键机制。基于这个视角，进一步阐述了可信赖人工智能的要求及人工智能治理的重要性，并分享了实验室在工业人工智能领域的实践及类脑通用人工智能的探索。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">北京智源人工智能研究院副院长兼总工程师林咏华分享了“众智FlagOS”的技术进展，指出该平台作为开放、统一的系统软件栈，旨在打破AI算力生态壁垒，实现跨地域、跨硬件的算力连接与普惠共享，为全球开发者提供跨芯片、跨框架、跨场景的统一计算底座。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">百度首席技术官王海峰梳理了人工智能从规则方法、统计机器学习到深度学习、大模型的发展历程，指出大模型技术通用性与能力的全面性为通用人工智能带来曙光。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">浪潮信息首席AI战略官刘军介绍了面向智能体时代的两大创新系统，分享了AI算力可持续发展面临的规模、电力、投入的挑战，提出要从规模导向转为效率导向，重新思考和设计AI计算系统，发展AI专用计算架构。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">北京中关村人工智能研究院副院长戴蓓洁介绍了北京中关村学院面向超常规AI领军人才培养需求，建立项目制人才培养体系，推动AI与多学科深度融合，实现科研成果落地与产业需求反哺技术创新的双向赋能，为新质生产力发展注入强劲动力。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong style=\"font-size: 18px; color: rgb(255, 153, 0);\" class=\"ql-lineHeight-1-75\">硬件创新瞄准token成本瓶颈\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">财联社记者现场采访了解到，在大会上，token成本高已成为众多企业推进AI应用规模化的核心痛点。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">“我们平台上每天有海量的客服、推荐、风控场景需要调用大模型，token成本就像悬在头上的‘达摩克利斯之剑’”。在大会现场，一位电商企业AI平台部的技术负责人对财联社记者透露，他正是为寻找“降本”方案而来。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">“随着智能体应用越铺越开，每个交互会话的token消耗量都在激增，目前的成本结构让很多有价值的创新应用在规模化落地前就卡在了‘经济账’上，盈利能力面临巨大考验。”上述负责人坦言，这也成为财联社记者在本届人工智能大会上听到的最具普遍性的声音之一。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">中国电子商务专家服务中心副主任郭涛在接受财联社记者采访时表示，当前AI产业正从“模型竞赛”转向“应用落地”，推理成本与交互速度已成为比模型参数规模更关键的竞争维度。基础设施的“提速降本”成效，将直接决定“人工智能+”在垂直行业渗透的深度与广度。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">会上，亦有一位上市公司人士对记者表示，随着Scaling Law持续推动模型能力跃升，以DeepSeek为代表的开源模型极大降低了创新门槛，加速智能体产业化的到来。智能体产业化的核心三要素是能力、速度和成本。其中，模型能力决定了智能体的应用上限，交互速度决定了智能体的商业价值，token成本决定了智能体的盈利能力。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">这一痛点在此次大会上引发广泛共鸣。面对业界的普遍诉求，算力基础设施厂商正试图从硬件层面寻求突破。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">硬件层面，浪潮信息在大会上发布元脑HC1000超扩展AI服务器，基于全新开发的全对称DirectCom极速架构，无损超扩展设计聚合海量本土AI芯片、支持极大推理吞吐量，推理成本首次击破1元\u002F每百万token，为智能体突破token成本瓶颈提供极致性能的创新算力系统。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">技术层面，刘军对财联社记者表示，元脑HC1000通过全面优化降本和软硬协同增效，创新16卡计算模组设计、单卡“计算-显存-互连”均衡设计，大幅降低单卡成本和每卡系统分摊成本。同时，全对称的系统拓扑设计支持超大规模无损扩展。据测算，元脑HC1000通过算网深度协同、全域无损技术，实现推理性能相比传统RoCE提升1.75倍，单卡模型算力利用率最高提升5.7倍。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">智能体未来带来的推理算力需求将呈现出指数级的爆发式增长，这一趋势无疑得到了行业的认同。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">浪潮信息方面对记者透露，将通过软硬件协同设计与深度优化，持续推动AI计算架构的创新与突破，不断实现token生成“提速降本”，积极促进大模型、智能体等人工智能技术与实体经济的深度融合，让人工智能成为千行百业的生产力和创新力。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"color: rgb(187, 187, 187);\">【新闻来源】财联社（记者 郭松峤） \u003C\u002Fspan>\u003Ca href=\"https:\u002F\u002Fwww.msn.cn\u002Fzh-cn\u002Fnews\u002Fother\u002Far-AA1NoO7k?ocid=msedgdhphdr&cvid=68d9e87721bd419981fe62e91b62d061&ei\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(187, 187, 187);\">https:\u002F\u002Fwww.msn.cn\u002Fzh-cn\u002Fnews\u002Fother\u002Far-AA1NoO7k?ocid=msedgdhphdr&cvid=68d9e87721bd419981fe62e91b62d061&ei\u003C\u002Fa>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(187, 187, 187);\">（本网转发此文章，旨在为读者提供更多的信息资讯，所涉内容不构成投资、消费建议。文章事实如有疑问，请与有关方核实，文章观点非本网观点，仅供读者参考。）\u003C\u002Fspan>\u003C\u002Fp>","","https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F09\u002F13c5e21d139749afb0f3d05667b1a014\u002FAI领域.jpg","https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F09\u002Fthumbs\u002F13c5e21d139749afb0f3d05667b1a014\u002FAI领域.jpg",0,1,47,"2025-09-29 19:08",2,false,{"id":17,"name":20,"enName":21},"芯位视野","Xinwei Vision","https:\u002F\u002Fxinwei-dev-test.oss-cn-shenzhen.aliyuncs.com\u002Fintelligent\u002Faudio%3A13797898-0ab0-4acb-b64d-334c2c99328a%3A0.wav?Expires=1759294005&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=YtHWopxvr%2B%2F02oUNY%2FvPL260c3g%3D",10407144,"13797898-0ab0-4acb-b64d-334c2c99328a","2025-09-29 19:06","2025 AI Computing Conference Observation: Token Cost Has Become a Bottleneck for the Large-Scale Application of AI, Industries Seek to Break Through the \"Impasse\"","\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">It has been just one month since the release of the \"Opinions on Deeply Implementing the 'AI+' Action,\" and the industry's \"start-up\" has already accelerated.\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The 2025 AI Computing Conference held in Beijing yesterday served as an excellent observation window. The reporter from Cailian Press noticed at the scene that the discussion focus of the participating experts and enterprise representatives had shifted from macro-level policy interpretation to specific \"construction blueprints.\"\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F09\u002F2b0181d29e774b09a170933f336d48f2\u002FAA1NoDZp.jpg\" width=\"556\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">This conference focused on the construction of artificial intelligence infrastructure and the optimization of the domestic AI computing system, focusing on promoting algorithm innovation and application implementation, using computing power as the core element to drive innovation, gathering forces from industry, academia, research, and application to jointly promote the high-quality development of the AI industry. At the site, more than 30 enterprises and institutions including China Mobile, Inspur Information, Zhiyuan Research Institute, and Kunlun Chip jointly released \"Building Industry Intelligent Bodies Based on Super Node Innovation Consortium - The Beijing Plan for Intelligent Computing Applications,\" responding to the national \"Opinions on Deeply Implementing the 'AI+' Action\" ahead of time.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong style=\"font-size: 18px; color: rgb(255, 153, 0);\" class=\"ql-lineHeight-1-75\">\"Achieving cross-regional and cross-hardware computing power connection and universal sharing\"\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">At the conference, Jin Yaochu, founder of the \"Trustworthy and General Artificial Intelligence Laboratory\" at West Lake University and academician of the European Academy of Sciences, outlined the main development trends of artificial intelligence and pointed out that its development path is similar to the emergence process of human brain intelligence, going through three key mechanisms: evolution, development, and learning. From this perspective, he further explained the requirements for trustworthy AI and the importance of AI governance, and shared the laboratory's practices in industrial AI and exploration of brain-like general AI.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Lin Yonghua, vice president and chief engineer of the Beijing Zhongguancun AI Institute, shared the technical progress of \"Zhongzhi FlagOS,\" pointing out that this platform, as an open and unified system software stack, aims to break down the barriers of the AI computing ecosystem, achieve cross-regional and cross-hardware computing power connection and universal sharing, and provide a unified computing base for global developers across chips, frameworks, and scenarios.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Wang Haifeng, Chief Technology Officer of Baidu, outlined the development process of artificial intelligence from rule-based methods, statistical machine learning, to deep learning and large models, pointing out that the universality and comprehensiveness of large model technology bring hope for general AI.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Liu Jun, Chief AI Strategy Officer of Inspur Information, introduced two innovative systems facing the era of intelligent agents, shared the challenges faced by sustainable development of AI computing in terms of scale, electricity, and investment, and proposed to shift from scale orientation to efficiency orientation, rethinking and designing AI computing systems, and developing AI-specific computing architectures.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Dai Beijie, vice president of the Beijing Zhongguancun AI Institute, introduced the project-based talent training system established by the Beijing Zhongguancun College to meet the demand for extraordinary AI leading talents, promoting the deep integration of AI with multiple disciplines, achieving the dual empowerment of scientific research results landing and industrial needs feeding back to technological innovation, and injecting strong momentum into the development of new quality productivity.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong style=\"font-size: 18px; color: rgb(255, 153, 0);\" class=\"ql-lineHeight-1-75\">Hardware innovation targets the token cost bottleneck\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">According to the on-site interview by Cailian Press reporters, token cost has become the core pain point for many companies to promote the large-scale application of AI at the conference.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">“Our platform has a massive number of customer service, recommendation, and risk control scenarios that require calling large models every day, and token costs are like the 'Sword of Damocles' hanging over our heads.” At the conference, a technical manager of an e-commerce company's AI platform revealed to Cailian Press reporters that he came here specifically to find a \"cost reduction\" solution.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">“As intelligent agent applications spread wider and wider, the token consumption per interaction is rapidly increasing. The current cost structure makes many valuable innovative applications face significant economic challenges before they can be scaled up. Profitability is facing huge tests,” the aforementioned official admitted, adding that this has also become one of the most common voices heard by Cailian Press reporters at this AI conference.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Guo Tao, deputy director of the China E-commerce Expert Service Center, told Cailian Press reporters that the AI industry is currently shifting from a \"model competition\" to \"application implementation.\" Inference cost and interaction speed have become more critical competitive dimensions than model parameter scale. The effectiveness of infrastructure \"speeding up and reducing costs\" will directly determine the depth and breadth of the penetration of \"AI+\" in vertical industries.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">At the conference, a person from a listed company told a reporter that as the Scaling Law continues to push the model capabilities to new heights, open-source models represented by DeepSeek have greatly reduced the innovation threshold, accelerating the arrival of intelligent agent industrialization. The three core elements of intelligent agent industrialization are capability, speed, and cost. Among them, model capability determines the application ceiling of intelligent agents, interaction speed determines the commercial value of intelligent agents, and token cost determines the profitability of intelligent agents.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">This pain point has triggered widespread resonance at this conference. In response to the widespread industry demands, computing infrastructure vendors are trying to seek breakthroughs from the hardware level.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">At the hardware level, Inspur Information released the Yuan Brain HC1000 super-expansible AI server at the conference, based on a newly developed fully symmetric DirectCom ultra-fast architecture, which features lossless super-expansive design, aggregating a massive amount of indigenous AI chips, supporting extremely high inference throughput, and for the first time breaking the 1 yuan per million tokens cost, providing an innovative computing system with extreme performance to help intelligent agents overcome the token cost bottleneck.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">In terms of technology, Liu Jun told Cailian Press reporters that the Yuan Brain HC1000 achieved comprehensive cost reduction through optimization and enhanced efficiency through soft-hardware collaboration, innovatively designed a 16-card computing module and balanced design of \"computing-memory-interconnect\" per card, significantly reducing the cost per card and the cost per card system allocation. At the same time, the fully symmetric system topology design supports ultra-large-scale lossless expansion. According to calculations, the Yuan Brain HC1000 achieves 1.75 times the inference performance compared to traditional RoCE, and the maximum improvement in single-card model computing power utilization is 5.7 times.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The inference computing power demand brought by intelligent agents will show exponential growth, a trend that has undoubtedly gained industry recognition.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Inspur Information informed reporters that it will continue to drive innovation and breakthroughs in AI computing architecture through collaborative design and in-depth optimization of hardware and software, constantly achieving \"speeding up and reducing costs\" for token generation, actively promoting the deep integration of AI technologies such as large models and intelligent agents with the real economy, making AI a productive force and innovative power in various industries.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"color: rgb(187, 187, 187);\">[News Source] Cailian Press (Reporter Guo Songqiao) \u003C\u002Fspan>\u003Ca href=\"https:\u002F\u002Fwww.msn.cn\u002Fzh-cn\u002Fnews\u002Fother\u002Far-AA1NoO7k?ocid=msedgdhphdr&cvid=68d9e87721bd419981fe62e91b62d061&ei\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(187, 187, 187);\">https:\u002F\u002Fwww.msn.cn\u002Fzh-cn\u002Fnews\u002Fother\u002Far-AA1NoO7k?ocid=msedgdhphdr&cvid=68d9e87721bd419981fe62e91b62d061&ei\u003C\u002Fa>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(187, 187, 187);\">（This article is reposted by this website to provide readers with more information and news. The content involved does not constitute investment or consumption advice. If there are any questions about the facts of the article, please verify with the relevant parties. The views expressed in the article are not the views of this website and are for reference only.）\u003C\u002Fspan>\u003C\u002Fp>","https:\u002F\u002Fxinwei-dev-test.oss-cn-shenzhen.aliyuncs.com\u002Fintelligent\u002Faudio%3Abbd0be29-b484-461c-8569-ef81a4479a6e%3A0.wav?Expires=1774838463&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=WkWxdhus4Bv6MCPewVVjS0IFMxk%3D","bbd0be29-b484-461c-8569-ef81a4479a6e",14401444]