[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fmEy0FNniK-531jB6L2Jm8nusUl77mdMiq7YjncSwO2k":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},1510,"视频生成大模型全球竞赛：2025年产业现状、六大品牌与未来走向","\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px; color: rgb(255, 153, 0);\">导语：可灵AI的月收入突破1亿人民币，其用户近一半来自海外市场。在全球排行榜上，中国企业的产品已占据了前十中的九席。\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">AI视频生成技术正在重新定义内容创作的边界。据《十五五视频行业发展研究与产业战略规划分析预测报告》，截至2025年第三季度末，人工智能在视频领域的渗透率已突破 \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">63%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">。市场监测数据显示，该领域的全球市场规模在2025年预计达到 \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">186亿美元\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">，较前一年增长近一倍。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">01 市场规模：产业拐点与爆发增长\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">当前视频生成产业正经历结构性变革。行业数据显示，AI驱动的视频生产成本较传统模式降低了 \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">47%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">，而用户的日均消费时长却同比增长了 \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">19%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">。产业拐点已经来临，资本与技术的双重驱动正加速这一进程。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">在商业应用层面，AI视频技术展现出强大的渗透力。统计显示，融合AI交互功能的社交平台日均用户停留时长是传统应用的 \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">2.8倍\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">。到2025年第三季度，头部视频社交平台中，\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">76%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> 的月活用户主动使用生成式内容创作功能。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">这种技术渗透直接转化为了商业价值。采用动态AI视频素材的品牌在电商场景中的点击率比静态图文高出 \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">41%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">，而单次点击成本则下降了 \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">19%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">。需求结构呈现出多元化的特点，从专业影视制作到个人社交媒体创作，从企业营销到教育娱乐，AI视频生成技术正在渗透到各个垂直领域。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">值得注意的是，\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">2025年全球视频相关投资规模较2024年增长了83%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">，其中智能生成工具占比31%，跨平台分发系统占27%，商业应用解决方案占42%。这些数据表明，产业正从单纯的技术竞争转向应用生态的全方位构建。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">02 全球格局：六大品牌形成中美双轨竞争\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">随着技术成熟和应用落地，全球视频生成大模型市场逐渐形成了清晰的竞争格局，呈现出中美双轨并行发展的态势。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">国外品牌方面\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">，OpenAI的Sora系列模型继续引领技术前沿，致力于探索基于一句提示词生成多镜头、角色一致且具有叙事连贯性的长视频。谷歌则通过Veo 3.0模型强化其AI电影制作工具Flow，并实现了视频与音频的原生集成与同步，代表了在多模态理解方面的深度探索。Runway作为较早进入该领域的公司，持续优化其创意工具集，在专业创作者中保持着重要地位。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">国内品牌方面\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">，竞争尤为激烈。快手可灵作为国内首个实现规模化商业落地的视频生成大模型，在2025年1月至5月的使用份额已超过 \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">30%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">，超越了Runway和Veo-2。字节跳动的即梦AI通过深度整合剪映工具链与抖音内容分发体系，形成了“创作-传播-变现”闭环。生数科技的Vidu则凭借其U-ViT融合架构，在画面真实感和细腻度上展现出独特优势。值得关注的是，根据中国报告大厅的报告，\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">当前全球排名前10的文生视频模型中，除谷歌外均由中国企业主导\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">。这反映出中国在AI视频应用领域的快速追赶和创新能力。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">03 技术特质：差异化路径构建核心壁垒\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">\t\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">各个品牌基于不同的技术路线和生态背景，形成了差异化的产品特质和竞争壁垒。\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">快手可灵\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">选择了与OpenAI Sora一致的DiT（Diffusion Transformer）架构，并在此基础上进行了多项自研创新。其核心技术包括3D VAE（变分自编码器）和3D时空联合注意力机制。3D VAE实现了时空同步压缩，使模型能够生成分辨率高达1080p、帧率达30fps的高质量视频。3D时空联合注意力机制则增强了模型对长期动态的建模能力，使其能够更好地理解视频中的复杂时空运动。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">近期，可灵进一步发布了视频O1模型，作为全球首个统一多模态视频大模型，打破了模态限制。用户可以通过自然语言对话，直接对视频进行内容增删、风格重绘等操作，使“P视频像P图一样简单”。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">字节即梦\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">的技术路线则以自研的Seedance 1.0系列模型为基础。该模型可生成多镜头无缝切换的1080p高品质视频，主体运动稳定性与画面自然度较高。通过统一的预训练框架，即梦提高了原生多镜头叙事能力，并实现了极致的推理加速，最快41秒就能生成5秒1080p的视频。即梦的核心优势在于其强大的生态整合能力。它将视频生成能力深度整合进剪映中，成为视频创作流程中的实用工具；生成的视频可以一键分享至抖音，对创作者作品起到宣传作用；同时还能与红果短剧等字节系产品深度配合，形成完整的“创作—剪辑—宣传—发行”商业化闭环。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">生数科技Vidu\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">则采用了独特的U-ViT架构，走了一条融合之路。在2025世界人工智能大会上，Vidu发布了“Vidu Q1参考生”功能，用户上传人物、道具、场景等参考图，就可以直接将多个参考元素生成为一段视频素材，以“参考图—视频生成—剪辑—成片”流程取代传统的分镜生成工作。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">从产品表现看，这三款国产模型已形成差异化特点。可灵优势在于表现力强，适合制作戏剧化内容；Vidu优势是真实、细腻，最有“电影感”；即梦则优势均衡、可控，工具属性突出。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">OpenAI Sora\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> 作为行业标杆，一直致力于探索长视频的叙事连贯性。它能够基于一句提示词生成多镜头、角色一致的长视频，展现了在视频理解和生成方面的强大能力。虽然尚未全面公测，但其技术路线和生成效果持续引领行业发展方向。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">谷歌Veo\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> 最大的突破在于实现了视频与音频的原生集成与同步，打破了AI视频的“无声尴尬”，划定了行业新标准。Veo 3模型展现了谷歌在多模态理解方面的深度探索，代表了视频生成从纯视觉向视听融合的重要演进。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">Runway\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> 作为较早进入该领域的公司，其优势在于创意工具集的完整性和对专业创作者工作流的深入理解。虽然面临后来者的激烈竞争，但Runway在特定创意领域仍保持着技术优势和市场地位。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">04 产业未来：从技术竞赛到生态构建\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">对于内容创作者、影视制作公司和营销机构等行业用户而言，选择国内视频生成大模型正在成为一个更加务实和高效的选择。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">国内模型通常采用更灵活的“免费+积分+低价会员”组合策略，降低了使用门槛。从效果看，国内头部产品在关键指标上已达到或接近国际先进水平，部分场景甚至表现更优。国内模型更贴近本土市场需求和创作习惯。中国拥有全球最庞大的互联网用户市场和极其活跃的内容创作生态，这为AI视频应用提供了绝佳的试验场和反馈池，推动技术在实践中快速迭代优化。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">以\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">快手可灵\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">为例，其商业化进展值得关注。今年3月，可灵的年化营收已突破 \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">1亿美元\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">，4月和5月的月度付费金额均超过 \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">1亿人民币\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">，营收增速和水平位居全球视频生成大模型产品前列。同时，可灵在全球创作者已超过 \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">4500万\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">，其中一大半是海外用户。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">\t\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">国内视频生成大模型正在从单纯的技术工具演变为\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">创作生态系统的重要组成部分\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">。随着技术的不断成熟和商业化路径的清晰，这些平台正在为创作者提供从灵感激发到作品分发的全链路支持。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">\t\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">行业数据显示，\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">AI驱动的视频营销项目平均投资回报率达到1:5.7，显著高于行业平均水平\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">。在直播带货领域，采用虚拟主播和实时特效的企业，其直播间观看完成率提升至68%，而行业均值为43%。随着可灵O1这类统一多模态模型的出现，视频创作的门槛将进一步降低，创意实现的效率将大幅提升。用户无需在多个工具间跳转，通过自然语言对话即可一站式完成从生成到修改的全部创作流程。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">\t\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">一家影视公司的制作人发现，使用国内AI工具生成广告素材，成本仅为传统制作的十分之一，且风格测试效率从每天几个方案提升到每小时18个。可灵AI在海外29个国家和地区的应用商店登上“图像和设计”类下载榜榜首。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"color: rgb(136, 136, 136);\">【新闻来源】凤凰网海南\u003C\u002Fspan>\u003Ca href=\"https:\u002F\u002Fmp.weixin.qq.com\u002Fs\u002Fmzzy7F-6NGpLDkMsy1w69g\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(136, 136, 136);\"> \u003C\u002Fa>\u003Ca href=\"https:\u002F\u002Fhainan.ifeng.com\u002Fc\u002F8omRbJ0RKNn\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(136, 136, 136); background-color: rgb(56, 56, 56); font-size: 14px;\">https:\u002F\u002Fhainan.ifeng.com\u002Fc\u002F8omRbJ0RKNn\u003C\u002Fa>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(136, 136, 136);\">（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>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>","","https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F12\u002F89d7272c65cf44ee8e3ecf2d5fb97ba0\u002Fd3ceb214-e320-4d27-b9ff-62bd5ab11fa0.jpg","https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F12\u002Fthumbs\u002F89d7272c65cf44ee8e3ecf2d5fb97ba0\u002Fd3ceb214-e320-4d27-b9ff-62bd5ab11fa0.jpg",0,1,48,"2025-12-15 10:15",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%3A2c1dbc95-ac13-4d2d-8f34-8d067b960665%3A0.wav?Expires=1765791684&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=2LgOZXQQ8oZxnRkBnfnZkFbIOmo%3D",16075348,"2c1dbc95-ac13-4d2d-8f34-8d067b960665","2025-12-12 10:06","Video Generation Large Model Global Competition: 2025 Industry Status, Six Major Brands, and Future Trends","\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px; color: rgb(255, 153, 0);\">Introduction: Keling AI's monthly revenue has exceeded 100 million RMB, with nearly half of its users coming from overseas markets. On the global ranking list, products from Chinese companies have occupied nine out of the top ten spots.\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">AI video generation technology is redefining the boundaries of content creation. According to the \"Fifteenth Five-Year Video Industry Development Research and Industrial Strategic Planning Analysis Forecast Report\", as of the end of the third quarter of 2025, the penetration rate of artificial intelligence in the video field has exceeded \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">63%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">. Market monitoring data shows that the global market size of this field is expected to reach \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">18.6 billion USD\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> in 2025, doubling compared to the previous year.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">01 Market Size: Industrial Turning Point and Explosive Growth\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">The current video generation industry is undergoing structural transformation. Industry data shows that the cost of AI-driven video production has been reduced by \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">47%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> compared to traditional models, while daily user consumption time has increased by \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">19%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">. The industrial turning point has arrived, and the dual drivers of capital and technology are accelerating this process.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">In commercial applications, AI video technology demonstrates strong penetration power. Statistics show that social platforms integrating AI interactive functions have an average user retention time 2.8 times that of traditional applications. By the third quarter of 2025, 76% of monthly active users on leading video social platforms actively use generative content creation features.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">This technological penetration directly translates into commercial value. Brands using dynamic AI video materials have a click-through rate 41% higher than static images and texts in e-commerce scenarios, while the cost per click has decreased by 19%. The demand structure shows diversified characteristics, ranging from professional film and television production to personal social media creation, from corporate marketing to education and entertainment. AI video generation technology is permeating various vertical fields.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">Notably, the global video-related investment scale in 2025 increased by 83% compared to 2024, with intelligent generation tools accounting for 31%, cross-platform distribution systems for 27%, and commercial application solutions for 42%. These data indicate that the industry is shifting from pure technological competition to comprehensive construction of application ecosystems.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">02 Global Landscape: Six Major Brands Forming a Sino-US Dual Track Competition\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">With the maturity of technology and the implementation of applications, the global video generation large model market has gradually formed a clear competitive landscape, showing a trend of parallel development between China and the United States.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">In terms of foreign brands\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">, OpenAI's Sora series model continues to lead the technological frontier, dedicated to exploring long videos with multiple scenes, consistent characters, and narrative coherence based on a single prompt. Google strengthens its AI movie-making tool Flow through the Veo 3.0 model and achieves native integration and synchronization of video and audio, representing a deep exploration in multimodal understanding. Runway, as one of the earlier companies entering this field, continues to optimize its creative toolset and maintains an important position among professional creators.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">In terms of domestic brands\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">, the competition is particularly fierce. Kuaishou Keling, as the first video generation large model in China to achieve large-scale commercial deployment, had a usage share exceeding \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">30%\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> from January to May 2025, surpassing Runway and Veo-2. ByteDance's Ji Meng AI integrates the Jianying toolchain and Douyin content distribution system deeply, forming a \"creation - dissemination - monetization\" closed loop. Shengshu Technology's Vidu showcases unique advantages in image realism and delicacy with its U-ViT fusion architecture. Notably, according to the report from China Report Hall,\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\"> all but one of the top 10 text-to-video models globally are led by Chinese companies\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">. This reflects China's rapid catch-up and innovation capabilities in the field of AI video applications.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">03 Technical Characteristics: Differentiation Pathways Building Core Barriers\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">\t\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">Based on different technical routes and ecological backgrounds, each brand has formed differentiated product characteristics and competitive barriers.\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">Kuaishou Keling\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> chose the DiT (Diffusion Transformer) architecture consistent with OpenAI Sora, and carried out several self-developed innovations on this basis. Its core technologies include 3D VAE (Variational Autoencoder) and 3D spatiotemporal joint attention mechanism. The 3D VAE achieves spatiotemporal synchronized compression, enabling the model to generate high-quality videos with a resolution of up to 1080p and a frame rate of 30fps. The 3D spatiotemporal joint attention mechanism enhances the model's ability to model long-term dynamics, allowing it to better understand complex spatiotemporal movements in videos.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">Recently, Keling further released the Video O1 model, as the world's first unified multimodal video large model, breaking the modal restrictions. Users can directly add or delete content and repaint the style of the video through natural language conversations, making \"P video as simple as P image.\"\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">ByteDance Ji Meng\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">'s technical route is based on its self-developed Seedance 1.0 series model. This model can generate high-quality 1080p videos with seamless multi-scene switching, with high stability of subject movement and naturalness of the image. Through a unified pre-training framework, Ji Meng improves the native multi-scene storytelling capability and achieves extreme reasoning acceleration, generating 5 seconds of 1080p video in as fast as 41 seconds. Ji Meng's core advantage lies in its powerful ecological integration capability. It deeply integrates video generation capabilities into Jianying, becoming a practical tool in the video creation process; the generated video can be shared to Douyin with one click, playing a promotional role for the creator's work; at the same time, it can also deeply cooperate with other ByteDance products such as Hongguo Short Plays, forming a complete \"creation - editing - promotion - distribution\" commercial closed loop.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">Shengshu Technology Vidu\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> adopted a unique U-ViT architecture, taking a path of integration. At the 2025 World Artificial Intelligence Conference, Vidu released the \"Vidu Q1 Reference Image\" function, allowing users to upload reference images of characters, props, and scenes, directly converting multiple reference elements into video materials, replacing the traditional storyboard generation workflow with a \"reference image - video generation - editing - final product\" process.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">From the product performance perspective, these three domestic models have formed differentiated characteristics. Keling excels in expressiveness, suitable for producing dramatic content; Vidu excels in realism and delicacy, having the most \"cinematic feel\"; Ji Meng has balanced advantages and controllability, with prominent tool attributes.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">OpenAI Sora\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">, as an industry benchmark, has always been dedicated to exploring the narrative coherence of long videos. It can generate long videos with multiple scenes and consistent characters based on a single prompt, demonstrating strong capabilities in video understanding and generation. Although not yet fully tested publicly, its technical route and generation effects continue to lead the direction of industry development.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">Google Veo\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> made a major breakthrough by achieving native integration and synchronization of video and audio, breaking the \"silent embarrassment\" of AI videos and setting a new industry standard. Veo 3 model demonstrated Google's deep exploration in multimodal understanding, representing an important evolution from pure visual to audiovisual integration in video generation.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">Runway\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">, as one of the earlier companies entering this field, has an advantage in the completeness of its creative toolset and a deep understanding of the workflow of professional creators. Although facing intense competition from later entrants, Runway still maintains technological advantages and market position in specific creative areas.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">04 Industry Future: From Technological Competition to Ecological Construction\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">For industry users such as content creators, film and television production companies, and marketing agencies, choosing domestic video generation large models has become a more practical and efficient choice.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">Domestic models usually adopt a more flexible \"free + points + low-cost membership\" combination strategy, lowering the usage threshold. In terms of effectiveness, the key indicators of the top domestic products have reached or approached international advanced levels, and in some scenarios, even performed better. Domestic models are more in line with local market demands and creation habits. China has the world's largest internet user market and an extremely active content creation ecosystem, providing an excellent test field and feedback pool for AI video applications, promoting rapid iteration and optimization of technology in practice.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">Taking\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">Kuaishou Keling\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> as an example, its commercial progress is worth noting. In March of this year, Keling's annualized revenue has exceeded \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">100 million USD\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">, and the monthly payment amount in April and May exceeded \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">100 million RMB\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">. Its revenue growth rate and level rank among the forefront of global video generation large model products. At the same time, Keling has over \u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">45 million\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\"> global creators, with nearly half being overseas users.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">\t\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">Domestic video generation large models are evolving from mere technological tools into\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">important components of the creative ecosystem\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">. With the continuous maturation of technology and the clarity of commercialization paths, these platforms are providing full-chain support for creators from inspiration to work distribution.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">\t\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">Industry data shows that\u003C\u002Fspan>\u003Cstrong style=\"font-size: 18px;\">AI-driven video marketing projects have an average return on investment of 1:5.7, significantly higher than the industry average\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">. In the live streaming e-commerce sector, companies using virtual anchors and real-time special effects have seen their live stream viewing completion rates rise to 68%, while the industry average is 43%. With the emergence of unified multimodal models like Keling O1, the threshold for video creation will be further lowered, and the efficiency of creative realization will be greatly improved. Users no longer need to switch between multiple tools, and they can complete the entire creative process from generation to modification through natural language conversation in one go.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">\t\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">A film and television production company found that using domestic AI tools to generate advertising materials costs only one-tenth of traditional production, and the efficiency of style testing increased from a few schemes per day to 18 per hour. Keling AI ranked first in the \"Images and Design\" download chart in application stores in 29 countries and regions abroad.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"color: rgb(136, 136, 136);\">【News Source】Phoenix TV Hainan\u003C\u002Fspan>\u003Ca href=\"https:\u002F\u002Fmp.weixin.qq.com\u002Fs\u002Fmzzy7F-6NGpLDkMsy1w69g\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(136, 136, 136);\"> \u003C\u002Fa>\u003Ca href=\"https:\u002F\u002Fhainan.ifeng.com\u002Fc\u002F8omRbJ0RKNn\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(136, 136, 136); background-color: rgb(56, 56, 56); font-size: 14px;\">https:\u002F\u002Fhainan.ifeng.com\u002Fc\u002F8omRbJ0RKNn\u003C\u002Fa>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(136, 136, 136);\">（This article is reposted by this site to provide readers with more information and news. The content does not constitute investment or consumer advice. If there are any doubts about the facts of the article, please verify with the relevant parties. The views of the article are not the views of this site and are for reference only.）\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>","https:\u002F\u002Fxinwei-dev-test.oss-cn-shenzhen.aliyuncs.com\u002Fintelligent\u002Faudio%3A08872d92-398c-4a4a-959e-40a7bf36c4a5%3A0.wav?Expires=1774838443&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=LPVlk3GY5RrRD8F0yK5PbJjQiVg%3D","08872d92-398c-4a4a-959e-40a7bf36c4a5",18263020]