{
  "text": "AI新闻不断涌现\n因为DeepSeek今天刚刚发布了全新的DeepSeek V4 FlashVision Experimental模型\n目前已在DeepSeek的API平台上线\n这个模型的多模态智能体性能已经接近OPUS 4.8\n确实是一个非常了不起的成就\n但唯一的阻碍是DeepSeek的新定价\n不过说到DeepSeek\n他们正在晖度测试一个神秘的新检查点\n即表现相当惊艳\n正在与Fibo 5正面交锋\n并生成了一些极其疯狂的内容\n不仅如此\nGoogle又回来了\n因为Geminis的表现真的会非常惊人\n具备一些疯狂的新能力\n我们已经获得了一些极其劲爆的泄露信息\n我非常期待大家看到这些内容\n最近我们报道了新的隐身模型OX-Alpha\n但现在关于这个模型又有了一些新情报\n其表现相当不可思议\n有人让它在DeepSway上完成了10项独立任务\n因此由于这只是部分测试\n它的真实得分可能存在很大差异\n但结果确实令人瞠目结舌\n因为GBT516S\n在这项特定测试中得分为52%\nFable 5得分为65%\n而Arx Alpha得分达到了80%\n据说它在其中一项任务上\n还差一点就完成了\n所以从技术上讲\n可能会超过80%\n这实在太疯狂了\n而且它真的感觉很像GLM模型\n但显然并不是\n因为有新的报告称\n它不是GLM模型\n它不是小米模型\n也不是Deep Seek模型\n所以没有人真正知道\n究竟是哪家模型提供商\n在秘密测试这个Arx Alpha模型\n但也许可能是minimax\n现在还有很多内容要介绍\n我们就直接一一深入了解吧\n大多数AI编程工具都能帮你构建应用\n但之后他们仍然需要处理托管\n数据库和部署问题\n这正是APICUS AI Supercomputer\n发挥作用的地方\n它提供了一台始终在线\n可通过自然语言控制的云端计算机\n你可以部署OpenCloud\n或Hermes Agent等智能体\n并让他们全天候持续运行\n例如他们在这里部署了一个\n开源大语言模型\n为其封装了一个可用的聊天界面\n并且仅通过一条提示词就将其公开托管\n你还可以运行Docker应用\n连接数据库和存储\n或者构建并托管完整的应用\n比如这个3D游戏\n因此你无需亲自管理VPS\nSSL 数据库和部署\nAppicus会处理整个技术站\n通过下方描述中的链接\n体验Appicus AI超级计算机\n也感谢他们赞助今天的视频\n我们先从Deepseek说起\n因为Deepseek Vs Flash Vision Experimental\n现已正式上线Deepseek API平台\n这个模型基本上就是VSFLAX的下一代多模态版本\n令人印象深刻的是\nDeepSeq表示\n这款新模型基本保留了VSFLAX的文本能力\n因此推理 智能体 编程和通用知识能力都还在\n但现在视觉方面得到了大幅升级\n从基准测试来看\n它的表现已经非常接近OPUS 4.8\n当然编程方面并不是这样\n但在多模态智能体评测方面却是如此\n在Terminal Bench 2.1上\n我们知道DeepSeq VSFLAX的得分为83.9\n而OPUS为85\n在DeepSWIG上\n上它实际上以59.3比58击败了OPUS\n但真正有趣的是多模态智能体评测\n在IpexBench上\n你可以看到这款新模型的得分为36.5\n而OPUS4.8的得分为39.4\n但在Agent's Last Exim上\n它的得分为27.3\n比OPUS4.8高出2分\n与OPUS4.8相比\n它在其他评测中的表现基本也只是略逊一筹\n因为它们实际上是在Version's Flash\n强大的视觉能力基础上\n进一步提升了多模态智能体的性能\n他们还已经在最新的Deep Seek Harness中加入了对它的支持\n因此你现在就可以通过他们的API开始测试这个模型\n而这正是视觉能力开始真正发挥作用的地方\n因为多模态能力能够拓展出更多智能体应用场景\nDeep Seek表示这款新模型可以在不同的智能体框架中流畅运行\n它现在既能理解自己看到的内容\n也能使用工具采取行动\n这样一来智能体就不再只能读取文本或代码\n而是可以查看屏幕截图\n界面 图表 文档或其他视觉信息\n理解实际发生的情况\n然后使用工具真正完成任务\n这让该模型在现实世界的智能体工作流中\n使用得多\n尤其适用于计算机操作\n视觉调试 浏览器智能体\n以及基于UI的文档自动化任务\n而且我们已经看到了一些\n相当令人印象深刻的演示\n展现了这次视觉能力升级\n究竟能做到什么\n有人立即通过给这个模型\n一张人物图片来测试它\n据说Deep Seek能够正确识别出图片中的人物\n是字节跳动创始人张一鸣\n显然这只是一个演示\n但这种说法相当有趣\nDeep Seek终于睁开了眼睛\n不再是盲人了\n更重要的是\n这表明Siflash Vision版本\n并不只是进行基本的图像描述\n而是能够理解详细的视觉向上闻\n并将其与已有知识结合起来\n无论是编程知识还是普通常识\n现在它将能够借助这一新扩展\n进一步增强自身能力\n如果你还不了解这一点\n我强烈建议你了解一下\nWorld of VI Insider School社区\n显然 在这里你可以获取大量信息\n不仅可以免费使用World of VI Benchmark\n还能每月获得许多免费内容\n甚至可以接触尚未发布的模型\n因此 我强烈建议你\n通过下方描述中的链接去了解一下\n不过说到Deep Seek\n其实还有一件更加有趣的事\n因为目前幕后正在发生的是\nDeep Seek团队正向一小部分用户\n大规模测试一个神秘的裸体模型检查点\n我之前已经针对这件事做过一期视频\n但现在又有一个新的检查点上线了\n它正凭借一些\n令人难以置信的生成结果\n与Fable 5展开激烈竞争\n有趣的是\n目前看来\n这种灰度测试\n似乎是基于绘画的测试\n也就是说\n某些现有绘画\n可能会被录游到\n这个实验性检查点\n而其他人则继续正常使用该模型\n现在一些用户\n甚至一直在尝试各种方法\n让自己的绘画\n持续被录游到灰度测试中\n不过显然\n这些都还无法说明\n这个检查点究竟是什么\n因此\nDeep Seek可能正在悄悄测试\n其下一代主力模型\n很可能是升级版的Version 4 Pro\n甚至是Version 5\n好了\n很抱歉画质不佳\n但这就是神秘的Deep Seek灰度检查点\n这里要求它在3D中建造一架直升机\n而它仅用一个提示词\n就完成了这项任务\n据说\n该模型自主运行了一个半小时\n持续构建并完善项目\n最终生成了这个细节惊人的\n直升机模拟程序\n你可以看到\n它在输出方面的能力有多么出色\n过程的成本还不到一美元\n这么说确实有点不可思议\n我想这个价格上的取舍\n大概就是生成内容所需的时间\n接下来我们终于看到了\n可能是GMMI4的首次亮相\n如果这些生成结果能够说明问题的话\n这款模型可能会强的离谱\n而Google也将凭借它真正实现王者归来\n这里要求它生成一只骑自行车的蹄壶\n这是一个相当不错的提示词\n我们通常总会把它发送给模型\n用来测试模型的能力\n显然主要是测试编程能力\n但你可以看到\n仅根据这段粉纹渲染\n它就生成了一只骑自行车的蹄壶\n效果非常惊艳\n你可以看到轮胎上\n甚至衣物上的文字印刷效果\n都生成得极其出色且清晰\n而且大多数模型实际上甚至做不到这一点\n连GML4模型在这种情况下\n所能完成的10%都不到\n但这简直太疯狂了\n如果GML4输出的质量能达到这种程度\n想象一下它在其他领域会有多强\n接下来看看AX Alpha\n我们已经知道这是一个隐秘模型\n而且表现相当出色\n对于错过相关消息的人来说\nOxelpha目前在接下来一周内都可以免费使用\n并且配备了100万Token的畅下源框口\n它支持多模态输入\n并且完全不保留数据\n显然它的速率限制也非常宽松\n每天最多可处理100万亿个Token\n但现在我们又获得了一些新情报\n让这个模型变得更加有趣\n有人让Oxelpha完成了10项Deeps Y任务\n不过这只是其中非常小的一部分\n因此结果可能存在很大差异\n这一点Dan也曾提到过\n但它的得分非常惊人\n达到了80%\n超过了得分65%的Febo\n以及得分52%的GBT 5.6 Soul\n据说\n其中一项实地任务\n基本上算是险些完成\n所以具体取决于你如何评分\n从技术上讲\n这个模型在这么小的样本规模下\n得分可能已经超过\n不过,我们之前还没提到的是,\n这并不只是一个轻量级的实验性端点。\n测试人员反映,\n它能够在超长上下纹下持续运行很长时间的智能体任务,\n这或许可以解释为什么它在软件工程任务上\n表现如此出色\n在这类任务中\n模型需要跟踪大量文件的状态\n以及之前做出的决策\n起初测试结果和社区论坛中的\n大量证据都指向同一个结论\nOxAlpha可能是某种新的GLM检查点\n但现在看来\n这一理论可能是错的\n因为有新的报告称\nOxAlpha并不是GLM\n也不是小米的Memo\n当然也不是DeepSeek的视觉模型\n不过事情在这里变得更加疯狂了\n因为还有传闻称\n该模型最终将以开放权重的形式发布\n据称它甚至可能在两台DGX Spark上本地运行\n现在谜团变得更大了\n如果它不是GLM 不是Memo 也不是DeepSeek\n我现在真的认为它可能是MiniX\n现在这是一个非常酷的演示\n展示了OX Alpha的多模态能力\n这次它运行在Hermes Agent中\n有人给它输入了一个类似Froggers的游戏提示词\nOX Alpha不仅构建了一个基础游戏\n还开始加入自己的创意机制\n它随机添加了一部会飞驰而来的ROG iPhone\n可以杀死青妈\n同时还在地图周围添加了青蛙可以收集的苍蝇\n以获得额外分数\n这正是多模态智能体如此有趣的地方\n模型可以构建游戏\n目视检查自己创建的内容\n理解并看到屏幕上实际发生了什么\n然后不断迭代\n直到找到最佳输出\n但看到这个模型竟然如此强大\n确实令人难以置信\n所以我非常想知道\n究竟是哪家AI实验室在专注于发布这个模型\n或者仍在对它进行测试\n如果你想在其他人之前\n获得最好的AI工具\n工作流和新产品动态\n请通过下方表述中的链接\n加入我的免费新闻简报\n完全免费\n有人让他创建一张\nQuincy Home的SVG图像\n希望我的发音没错\n内容是他在北方巡逻期间\n骑着一头北极熊\n而且这实际上是我见过的\n所有中国AI模型中\n生成效果最好的SVG之一\n它确实能够让许多功能\n实现动画效果\n所以看到他能够做到这样的事情\n真的很酷\n接下来看看Claude\nAnthropic刚刚进行了一次\n相当重大的智能体更新\n因为Computer Use 浏览器工具\nSkills API和Files API\n现在已在Cloud平台上全面开放\n这里的核心理念是\nCloud的智能体\n现在不只是调用普通API了\n借助Computer Use\nCloud可以与甚至没有API的应用\n进行交互\n而浏览器工具\n则让它能够直接在整个网络中\n导航和执行操作\n然后还有Skills API\n它允许开发者为Cloud的智能体\n提供可复用的技能和能力\nFiles API则会让智能体\n更轻松的在不同任务中处理同一批文档和文件\n因此\nAnthropic显然正致力于打造一个完整的云端托管智能体系统\n使其能够真正完成从Computer Use\n浏览器工具到Skills API和Files API等各种不同的操作\n如果你喜欢这段视频并愿意支持本频道\n可以考虑通过下方的超级感谢选项向我的频道捐赠\n或者你也可以考虑加入我们的私人Discord服务器\n在那里\n你每月都可以免费使用多种AI工具的订阅服务\n还能获取每日AI新闻\n独家内容\n以及更多福利\n或许如果我有时间\n我可能会专门制作一期完整视频\n来介绍ONEK 1.5\n因为这确实是一个非常出色的本地模型\n能力相当强\n显然它在某些方面\n无法与Qing 3.8 Max相提并论\n但你仍然可以从中获得很多成果\n它在不少任务上都表现得相当不错\n当然使用更低的位数时\n输出质量也会有所下降\n如果你实际运行这个模型的\n350亿参数5Q版本\n它能够为你带来一些\n令人难以置信的结果\n临时需要马上补充编码的新闻\n据报道\nChatJPT Work和Codex用户\n今天似乎即将演来一个相当不错的惊喜\n因为Table今天宣布了一项\n以内置的使用额度重置功能\n而这一消息发布之际\nCodex的活跃用户数已经突破2000万\n这确实相当惊人\n本质上\n这将为用户提供一批全新的Codex使用额度\n供他们消耗\n因此\n你的速率限制将恢复正常\n或者说恢复到100%\n这意味着\n你可以在整个周末继续使用Codex\n不过今天的视频就到这里了\n我会把今天视频中使用的所有链接\n放在下方的描述栏里\n记得去看看World of AI Insider Club School社区\n记得订阅第二频道\n加入新闻通讯\n加入Discord\n在Twitter上关注我\n最后记得订阅频道\n打开通知铃\n给这个视频点赞\n也请看看我们之前的视频\n以便及时了解最新的AI新闻\n说到这里\n朋友们\n祝大家度过美好的一天\n传播积极正能量\n我们很快再见\n愿他受苦",
  "language": "multilingual",
  "duration": 754.112,
  "segments": [
    {
      "start": 0.0,
      "end": 1.38,
      "text": "AI新闻不断涌现",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 1.38,
      "end": 6.42,
      "text": "因为DeepSeek今天刚刚发布了全新的DeepSeek V4 FlashVision Experimental模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 6.42,
      "end": 8.82,
      "text": "目前已在DeepSeek的API平台上线",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 8.82,
      "end": 12.74,
      "text": "这个模型的多模态智能体性能已经接近OPUS 4.8",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 12.74,
      "end": 14.92,
      "text": "确实是一个非常了不起的成就",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 14.92,
      "end": 17.66,
      "text": "但唯一的阻碍是DeepSeek的新定价",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 17.66,
      "end": 18.72,
      "text": "不过说到DeepSeek",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 18.72,
      "end": 21.5,
      "text": "他们正在晖度测试一个神秘的新检查点",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 21.5,
      "end": 23.06,
      "text": "即表现相当惊艳",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 23.06,
      "end": 24.84,
      "text": "正在与Fibo 5正面交锋",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 24.84,
      "end": 27.02,
      "text": "并生成了一些极其疯狂的内容",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 27.02,
      "end": 27.54,
      "text": "不仅如此",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 27.54,
      "end": 28.6,
      "text": "Google又回来了",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 28.6,
      "end": 31.3,
      "text": "因为Geminis的表现真的会非常惊人",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 31.3,
      "end": 32.9,
      "text": "具备一些疯狂的新能力",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 32.9,
      "end": 35.46,
      "text": "我们已经获得了一些极其劲爆的泄露信息",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 35.46,
      "end": 37.88,
      "text": "我非常期待大家看到这些内容",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 37.88,
      "end": 40.46,
      "text": "最近我们报道了新的隐身模型OX-Alpha",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 40.46,
      "end": 42.9,
      "text": "但现在关于这个模型又有了一些新情报",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 42.9,
      "end": 44.76,
      "text": "其表现相当不可思议",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 44.76,
      "end": 47.76,
      "text": "有人让它在DeepSway上完成了10项独立任务",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 47.76,
      "end": 49.72,
      "text": "因此由于这只是部分测试",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 49.72,
      "end": 52.08,
      "text": "它的真实得分可能存在很大差异",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 52.08,
      "end": 54.0,
      "text": "但结果确实令人瞠目结舌",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 54.0,
      "end": 56.0,
      "text": "因为GBT516S",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 56.0,
      "end": 58.68,
      "text": "在这项特定测试中得分为52%",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 58.68,
      "end": 60.92,
      "text": "Fable 5得分为65%",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 60.92,
      "end": 63.88,
      "text": "而Arx Alpha得分达到了80%",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 63.88,
      "end": 65.64,
      "text": "据说它在其中一项任务上",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 65.64,
      "end": 67.12,
      "text": "还差一点就完成了",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 67.12,
      "end": 68.42,
      "text": "所以从技术上讲",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 68.42,
      "end": 70.06,
      "text": "可能会超过80%",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 70.06,
      "end": 71.36,
      "text": "这实在太疯狂了",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 71.36,
      "end": 73.8,
      "text": "而且它真的感觉很像GLM模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 73.8,
      "end": 74.98,
      "text": "但显然并不是",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 74.98,
      "end": 76.32,
      "text": "因为有新的报告称",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 76.32,
      "end": 77.82,
      "text": "它不是GLM模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 77.82,
      "end": 79.1,
      "text": "它不是小米模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 79.1,
      "end": 80.36,
      "text": "也不是Deep Seek模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 80.36,
      "end": 82.12,
      "text": "所以没有人真正知道",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 82.12,
      "end": 83.82,
      "text": "究竟是哪家模型提供商",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 83.82,
      "end": 85.72,
      "text": "在秘密测试这个Arx Alpha模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 85.72,
      "end": 87.44,
      "text": "但也许可能是minimax",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 87.44,
      "end": 89.12,
      "text": "现在还有很多内容要介绍",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 89.12,
      "end": 90.98,
      "text": "我们就直接一一深入了解吧",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 90.98,
      "end": 93.82,
      "text": "大多数AI编程工具都能帮你构建应用",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 93.82,
      "end": 95.98,
      "text": "但之后他们仍然需要处理托管",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 95.98,
      "end": 97.28,
      "text": "数据库和部署问题",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 97.28,
      "end": 99.54,
      "text": "这正是APICUS AI Supercomputer",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 99.54,
      "end": 100.52,
      "text": "发挥作用的地方",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 100.52,
      "end": 102.42,
      "text": "它提供了一台始终在线",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 102.42,
      "end": 104.84,
      "text": "可通过自然语言控制的云端计算机",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 104.84,
      "end": 106.32,
      "text": "你可以部署OpenCloud",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 106.32,
      "end": 108.18,
      "text": "或Hermes Agent等智能体",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 108.18,
      "end": 110.26,
      "text": "并让他们全天候持续运行",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 110.26,
      "end": 111.92,
      "text": "例如他们在这里部署了一个",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 111.92,
      "end": 113.2,
      "text": "开源大语言模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 113.2,
      "end": 115.4,
      "text": "为其封装了一个可用的聊天界面",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 115.4,
      "end": 118.82,
      "text": "并且仅通过一条提示词就将其公开托管",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 118.82,
      "end": 120.44,
      "text": "你还可以运行Docker应用",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 120.44,
      "end": 122.0,
      "text": "连接数据库和存储",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 122.0,
      "end": 124.32,
      "text": "或者构建并托管完整的应用",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 124.32,
      "end": 125.74,
      "text": "比如这个3D游戏",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 125.74,
      "end": 127.82,
      "text": "因此你无需亲自管理VPS",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 127.82,
      "end": 129.86,
      "text": "SSL 数据库和部署",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 129.86,
      "end": 131.66,
      "text": "Appicus会处理整个技术站",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 131.66,
      "end": 133.22,
      "text": "通过下方描述中的链接",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 133.22,
      "end": 135.26,
      "text": "体验Appicus AI超级计算机",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 135.26,
      "end": 137.28,
      "text": "也感谢他们赞助今天的视频",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 137.28,
      "end": 138.86,
      "text": "我们先从Deepseek说起",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 138.86,
      "end": 141.24,
      "text": "因为Deepseek Vs Flash Vision Experimental",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 141.24,
      "end": 143.74,
      "text": "现已正式上线Deepseek API平台",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 143.74,
      "end": 147.42,
      "text": "这个模型基本上就是VSFLAX的下一代多模态版本",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 147.42,
      "end": 148.52,
      "text": "令人印象深刻的是",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 148.52,
      "end": 149.44,
      "text": "DeepSeq表示",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 149.44,
      "end": 152.58,
      "text": "这款新模型基本保留了VSFLAX的文本能力",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 152.58,
      "end": 156.34,
      "text": "因此推理 智能体 编程和通用知识能力都还在",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 156.34,
      "end": 159.18,
      "text": "但现在视觉方面得到了大幅升级",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 159.18,
      "end": 160.22,
      "text": "从基准测试来看",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 160.22,
      "end": 162.6,
      "text": "它的表现已经非常接近OPUS 4.8",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 162.6,
      "end": 164.32,
      "text": "当然编程方面并不是这样",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 164.32,
      "end": 167.2,
      "text": "但在多模态智能体评测方面却是如此",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 167.2,
      "end": 169.1,
      "text": "在Terminal Bench 2.1上",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 169.1,
      "end": 172.2,
      "text": "我们知道DeepSeq VSFLAX的得分为83.9",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 172.2,
      "end": 173.6,
      "text": "而OPUS为85",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 173.6,
      "end": 174.78,
      "text": "在DeepSWIG上",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 174.78,
      "end": 178.32,
      "text": "上它实际上以59.3比58击败了OPUS",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 178.32,
      "end": 181.08,
      "text": "但真正有趣的是多模态智能体评测",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 181.08,
      "end": 182.38,
      "text": "在IpexBench上",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 182.38,
      "end": 185.14,
      "text": "你可以看到这款新模型的得分为36.5",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 185.14,
      "end": 187.94,
      "text": "而OPUS4.8的得分为39.4",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 187.94,
      "end": 189.78,
      "text": "但在Agent's Last Exim上",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 189.78,
      "end": 191.5,
      "text": "它的得分为27.3",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 191.5,
      "end": 193.64,
      "text": "比OPUS4.8高出2分",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 193.64,
      "end": 195.12,
      "text": "与OPUS4.8相比",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 195.12,
      "end": 198.22,
      "text": "它在其他评测中的表现基本也只是略逊一筹",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 198.22,
      "end": 200.36,
      "text": "因为它们实际上是在Version's Flash",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 200.36,
      "end": 201.96,
      "text": "强大的视觉能力基础上",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 201.96,
      "end": 204.8,
      "text": "进一步提升了多模态智能体的性能",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 204.8,
      "end": 208.2,
      "text": "他们还已经在最新的Deep Seek Harness中加入了对它的支持",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 208.2,
      "end": 212.24,
      "text": "因此你现在就可以通过他们的API开始测试这个模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 212.24,
      "end": 215.34,
      "text": "而这正是视觉能力开始真正发挥作用的地方",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 215.34,
      "end": 219.32,
      "text": "因为多模态能力能够拓展出更多智能体应用场景",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 219.32,
      "end": 223.62,
      "text": "Deep Seek表示这款新模型可以在不同的智能体框架中流畅运行",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 223.62,
      "end": 225.82,
      "text": "它现在既能理解自己看到的内容",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 225.82,
      "end": 227.98,
      "text": "也能使用工具采取行动",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 227.98,
      "end": 230.92,
      "text": "这样一来智能体就不再只能读取文本或代码",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 230.92,
      "end": 232.66,
      "text": "而是可以查看屏幕截图",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 232.66,
      "end": 235.62,
      "text": "界面 图表 文档或其他视觉信息",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 235.62,
      "end": 237.16,
      "text": "理解实际发生的情况",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 237.16,
      "end": 239.44,
      "text": "然后使用工具真正完成任务",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 239.44,
      "end": 242.18,
      "text": "这让该模型在现实世界的智能体工作流中",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 242.18,
      "end": 242.82,
      "text": "使用得多",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 242.82,
      "end": 244.56,
      "text": "尤其适用于计算机操作",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 244.56,
      "end": 246.5,
      "text": "视觉调试 浏览器智能体",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 246.5,
      "end": 249.18,
      "text": "以及基于UI的文档自动化任务",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 249.18,
      "end": 250.7,
      "text": "而且我们已经看到了一些",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 250.7,
      "end": 252.52,
      "text": "相当令人印象深刻的演示",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 252.52,
      "end": 254.38,
      "text": "展现了这次视觉能力升级",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 254.38,
      "end": 255.48,
      "text": "究竟能做到什么",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 255.48,
      "end": 257.24,
      "text": "有人立即通过给这个模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 257.24,
      "end": 258.74,
      "text": "一张人物图片来测试它",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 258.74,
      "end": 262.08,
      "text": "据说Deep Seek能够正确识别出图片中的人物",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 262.08,
      "end": 264.14,
      "text": "是字节跳动创始人张一鸣",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 264.14,
      "end": 265.82,
      "text": "显然这只是一个演示",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 265.82,
      "end": 267.5,
      "text": "但这种说法相当有趣",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 267.5,
      "end": 269.5,
      "text": "Deep Seek终于睁开了眼睛",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 269.5,
      "end": 270.74,
      "text": "不再是盲人了",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 270.74,
      "end": 271.66,
      "text": "更重要的是",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 271.66,
      "end": 273.52,
      "text": "这表明Siflash Vision版本",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 273.52,
      "end": 275.62,
      "text": "并不只是进行基本的图像描述",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 275.62,
      "end": 278.06,
      "text": "而是能够理解详细的视觉向上闻",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 278.06,
      "end": 280.3,
      "text": "并将其与已有知识结合起来",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 280.3,
      "end": 282.2,
      "text": "无论是编程知识还是普通常识",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 282.2,
      "end": 284.5,
      "text": "现在它将能够借助这一新扩展",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 284.5,
      "end": 286.0,
      "text": "进一步增强自身能力",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 286.0,
      "end": 287.44,
      "text": "如果你还不了解这一点",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 287.44,
      "end": 289.16,
      "text": "我强烈建议你了解一下",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 289.16,
      "end": 290.8,
      "text": "World of VI Insider School社区",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 290.8,
      "end": 293.44,
      "text": "显然 在这里你可以获取大量信息",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 293.44,
      "end": 295.74,
      "text": "不仅可以免费使用World of VI Benchmark",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 295.74,
      "end": 297.86,
      "text": "还能每月获得许多免费内容",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 297.86,
      "end": 300.44,
      "text": "甚至可以接触尚未发布的模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 300.44,
      "end": 301.94,
      "text": "因此 我强烈建议你",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 301.94,
      "end": 304.18,
      "text": "通过下方描述中的链接去了解一下",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 304.18,
      "end": 305.38,
      "text": "不过说到Deep Seek",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 305.38,
      "end": 307.1,
      "text": "其实还有一件更加有趣的事",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 307.1,
      "end": 308.94,
      "text": "因为目前幕后正在发生的是",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 308.94,
      "end": 311.26,
      "text": "Deep Seek团队正向一小部分用户",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 311.26,
      "end": 314.38,
      "text": "大规模测试一个神秘的裸体模型检查点",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 314.38,
      "end": 316.62,
      "text": "我之前已经针对这件事做过一期视频",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 316.62,
      "end": 319.06,
      "text": "但现在又有一个新的检查点上线了",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 319.06,
      "end": 320.02,
      "text": "它正凭借一些",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 320.02,
      "end": 321.76,
      "text": "令人难以置信的生成结果",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 321.76,
      "end": 323.82,
      "text": "与Fable 5展开激烈竞争",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 323.82,
      "end": 324.42,
      "text": "有趣的是",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 324.42,
      "end": 325.32,
      "text": "目前看来",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 325.32,
      "end": 326.46,
      "text": "这种灰度测试",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 326.46,
      "end": 328.46,
      "text": "似乎是基于绘画的测试",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 328.46,
      "end": 329.24,
      "text": "也就是说",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 329.24,
      "end": 330.5,
      "text": "某些现有绘画",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 330.5,
      "end": 331.72,
      "text": "可能会被录游到",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 331.72,
      "end": 332.94,
      "text": "这个实验性检查点",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 332.94,
      "end": 335.48,
      "text": "而其他人则继续正常使用该模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 335.48,
      "end": 336.66,
      "text": "现在一些用户",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 336.66,
      "end": 338.42,
      "text": "甚至一直在尝试各种方法",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 338.42,
      "end": 339.5,
      "text": "让自己的绘画",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 339.5,
      "end": 341.44,
      "text": "持续被录游到灰度测试中",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 341.44,
      "end": 342.16,
      "text": "不过显然",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 342.16,
      "end": 343.56,
      "text": "这些都还无法说明",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 343.56,
      "end": 345.02,
      "text": "这个检查点究竟是什么",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 345.02,
      "end": 345.72,
      "text": "因此",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 345.72,
      "end": 347.68,
      "text": "Deep Seek可能正在悄悄测试",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 347.68,
      "end": 349.04,
      "text": "其下一代主力模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 349.04,
      "end": 351.16,
      "text": "很可能是升级版的Version 4 Pro",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 351.16,
      "end": 352.72,
      "text": "甚至是Version 5",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 352.72,
      "end": 352.9,
      "text": "好了",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 352.9,
      "end": 354.2,
      "text": "很抱歉画质不佳",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 354.2,
      "end": 356.64,
      "text": "但这就是神秘的Deep Seek灰度检查点",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 356.64,
      "end": 359.22,
      "text": "这里要求它在3D中建造一架直升机",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 359.22,
      "end": 360.76,
      "text": "而它仅用一个提示词",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 360.76,
      "end": 362.24,
      "text": "就完成了这项任务",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 362.24,
      "end": 362.58,
      "text": "据说",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 362.58,
      "end": 364.6,
      "text": "该模型自主运行了一个半小时",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 364.6,
      "end": 366.5,
      "text": "持续构建并完善项目",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 366.5,
      "end": 368.58,
      "text": "最终生成了这个细节惊人的",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 368.58,
      "end": 370.04,
      "text": "直升机模拟程序",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 370.04,
      "end": 370.6,
      "text": "你可以看到",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 370.6,
      "end": 372.7,
      "text": "它在输出方面的能力有多么出色",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 373.56,
      "end": 375.18,
      "text": "过程的成本还不到一美元",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 375.18,
      "end": 377.04,
      "text": "这么说确实有点不可思议",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 377.04,
      "end": 378.96,
      "text": "我想这个价格上的取舍",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 378.96,
      "end": 381.02,
      "text": "大概就是生成内容所需的时间",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 381.02,
      "end": 383.1,
      "text": "接下来我们终于看到了",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 383.1,
      "end": 385.32,
      "text": "可能是GMMI4的首次亮相",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 385.32,
      "end": 387.62,
      "text": "如果这些生成结果能够说明问题的话",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 387.62,
      "end": 389.58,
      "text": "这款模型可能会强的离谱",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 389.58,
      "end": 392.76,
      "text": "而Google也将凭借它真正实现王者归来",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 392.76,
      "end": 395.52,
      "text": "这里要求它生成一只骑自行车的蹄壶",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 395.52,
      "end": 397.28,
      "text": "这是一个相当不错的提示词",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 397.28,
      "end": 399.3,
      "text": "我们通常总会把它发送给模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 399.3,
      "end": 400.64,
      "text": "用来测试模型的能力",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 400.64,
      "end": 402.5,
      "text": "显然主要是测试编程能力",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 402.5,
      "end": 403.32,
      "text": "但你可以看到",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 403.32,
      "end": 404.9,
      "text": "仅根据这段粉纹渲染",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 404.9,
      "end": 407.08,
      "text": "它就生成了一只骑自行车的蹄壶",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 407.08,
      "end": 408.6,
      "text": "效果非常惊艳",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 408.6,
      "end": 410.06,
      "text": "你可以看到轮胎上",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 410.06,
      "end": 411.96,
      "text": "甚至衣物上的文字印刷效果",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 411.96,
      "end": 414.42,
      "text": "都生成得极其出色且清晰",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 414.42,
      "end": 416.78,
      "text": "而且大多数模型实际上甚至做不到这一点",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 416.78,
      "end": 419.04,
      "text": "连GML4模型在这种情况下",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 419.04,
      "end": 420.82,
      "text": "所能完成的10%都不到",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 420.82,
      "end": 422.44,
      "text": "但这简直太疯狂了",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 422.44,
      "end": 425.14,
      "text": "如果GML4输出的质量能达到这种程度",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 425.14,
      "end": 427.38,
      "text": "想象一下它在其他领域会有多强",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 427.38,
      "end": 428.76,
      "text": "接下来看看AX Alpha",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 428.76,
      "end": 430.72,
      "text": "我们已经知道这是一个隐秘模型",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 430.72,
      "end": 432.18,
      "text": "而且表现相当出色",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 432.18,
      "end": 434.08,
      "text": "对于错过相关消息的人来说",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 434.08,
      "end": 437.14,
      "text": "Oxelpha目前在接下来一周内都可以免费使用",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 437.14,
      "end": 440.24,
      "text": "并且配备了100万Token的畅下源框口",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 440.24,
      "end": 441.68,
      "text": "它支持多模态输入",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 441.68,
      "end": 443.5,
      "text": "并且完全不保留数据",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 443.5,
      "end": 445.94,
      "text": "显然它的速率限制也非常宽松",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 445.94,
      "end": 448.56,
      "text": "每天最多可处理100万亿个Token",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 448.56,
      "end": 450.66,
      "text": "但现在我们又获得了一些新情报",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 450.66,
      "end": 452.54,
      "text": "让这个模型变得更加有趣",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 452.54,
      "end": 455.88,
      "text": "有人让Oxelpha完成了10项Deeps Y任务",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 455.88,
      "end": 457.94,
      "text": "不过这只是其中非常小的一部分",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 457.94,
      "end": 460.16,
      "text": "因此结果可能存在很大差异",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 460.16,
      "end": 461.48,
      "text": "这一点Dan也曾提到过",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 461.48,
      "end": 463.22,
      "text": "但它的得分非常惊人",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 463.22,
      "end": 464.62,
      "text": "达到了80%",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 464.62,
      "end": 466.76,
      "text": "超过了得分65%的Febo",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 466.76,
      "end": 469.5,
      "text": "以及得分52%的GBT 5.6 Soul",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 469.5,
      "end": 470.12,
      "text": "据说",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 470.12,
      "end": 471.48,
      "text": "其中一项实地任务",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 471.48,
      "end": 472.98,
      "text": "基本上算是险些完成",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 472.98,
      "end": 475.6,
      "text": "所以具体取决于你如何评分",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 475.6,
      "end": 476.58,
      "text": "从技术上讲",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 476.58,
      "end": 478.68,
      "text": "这个模型在这么小的样本规模下",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 478.68,
      "end": 480.0,
      "text": "得分可能已经超过",
      "language": "zh",
      "chunk": 0,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 481.1,
      "end": 482.74,
      "text": "不过,我们之前还没提到的是,",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 483.04,
      "end": 485.38,
      "text": "这并不只是一个轻量级的实验性端点。",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 485.78,
      "end": 486.52,
      "text": "测试人员反映,",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 486.52,
      "end": 491.64,
      "text": "它能够在超长上下纹下持续运行很长时间的智能体任务,",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 491.84,
      "end": 494.72,
      "text": "这或许可以解释为什么它在软件工程任务上",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 494.72,
      "end": 495.9,
      "text": "表现如此出色",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 495.9,
      "end": 497.08,
      "text": "在这类任务中",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 497.08,
      "end": 499.44,
      "text": "模型需要跟踪大量文件的状态",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 499.44,
      "end": 501.28,
      "text": "以及之前做出的决策",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 501.28,
      "end": 503.46,
      "text": "起初测试结果和社区论坛中的",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 503.46,
      "end": 505.22,
      "text": "大量证据都指向同一个结论",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 505.22,
      "end": 508.7,
      "text": "OxAlpha可能是某种新的GLM检查点",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 508.7,
      "end": 509.7,
      "text": "但现在看来",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 509.7,
      "end": 511.16,
      "text": "这一理论可能是错的",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 511.16,
      "end": 512.34,
      "text": "因为有新的报告称",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 512.34,
      "end": 514.18,
      "text": "OxAlpha并不是GLM",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 514.18,
      "end": 515.48,
      "text": "也不是小米的Memo",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 515.48,
      "end": 517.48,
      "text": "当然也不是DeepSeek的视觉模型",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 517.48,
      "end": 520.1,
      "text": "不过事情在这里变得更加疯狂了",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 520.1,
      "end": 521.16,
      "text": "因为还有传闻称",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 521.16,
      "end": 524.12,
      "text": "该模型最终将以开放权重的形式发布",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 524.12,
      "end": 527.72,
      "text": "据称它甚至可能在两台DGX Spark上本地运行",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 527.72,
      "end": 529.74,
      "text": "现在谜团变得更大了",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 529.74,
      "end": 532.98,
      "text": "如果它不是GLM 不是Memo 也不是DeepSeek",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 532.98,
      "end": 535.02,
      "text": "我现在真的认为它可能是MiniX",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 535.02,
      "end": 537.38,
      "text": "现在这是一个非常酷的演示",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 537.38,
      "end": 539.74,
      "text": "展示了OX Alpha的多模态能力",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 539.74,
      "end": 541.64,
      "text": "这次它运行在Hermes Agent中",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 541.64,
      "end": 544.58,
      "text": "有人给它输入了一个类似Froggers的游戏提示词",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 544.58,
      "end": 547.08,
      "text": "OX Alpha不仅构建了一个基础游戏",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 547.08,
      "end": 549.12,
      "text": "还开始加入自己的创意机制",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 549.12,
      "end": 551.94,
      "text": "它随机添加了一部会飞驰而来的ROG iPhone",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 551.94,
      "end": 552.92,
      "text": "可以杀死青妈",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 552.92,
      "end": 556.12,
      "text": "同时还在地图周围添加了青蛙可以收集的苍蝇",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 556.12,
      "end": 557.36,
      "text": "以获得额外分数",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 557.36,
      "end": 560.02,
      "text": "这正是多模态智能体如此有趣的地方",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 560.02,
      "end": 561.56,
      "text": "模型可以构建游戏",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 561.56,
      "end": 563.44,
      "text": "目视检查自己创建的内容",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 563.44,
      "end": 565.98,
      "text": "理解并看到屏幕上实际发生了什么",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 565.98,
      "end": 567.2,
      "text": "然后不断迭代",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 567.2,
      "end": 568.88,
      "text": "直到找到最佳输出",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 568.88,
      "end": 570.78,
      "text": "但看到这个模型竟然如此强大",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 570.78,
      "end": 572.12,
      "text": "确实令人难以置信",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 572.12,
      "end": 573.38,
      "text": "所以我非常想知道",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 573.38,
      "end": 576.82,
      "text": "究竟是哪家AI实验室在专注于发布这个模型",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 576.82,
      "end": 578.72,
      "text": "或者仍在对它进行测试",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 578.72,
      "end": 580.3,
      "text": "如果你想在其他人之前",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 580.3,
      "end": 581.76,
      "text": "获得最好的AI工具",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 581.76,
      "end": 583.48,
      "text": "工作流和新产品动态",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 583.48,
      "end": 585.32,
      "text": "请通过下方表述中的链接",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 585.32,
      "end": 587.0,
      "text": "加入我的免费新闻简报",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 587.0,
      "end": 587.82,
      "text": "完全免费",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 587.82,
      "end": 589.18,
      "text": "有人让他创建一张",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 589.18,
      "end": 590.86,
      "text": "Quincy Home的SVG图像",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 590.86,
      "end": 591.9,
      "text": "希望我的发音没错",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 591.9,
      "end": 594.08,
      "text": "内容是他在北方巡逻期间",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 594.08,
      "end": 595.48,
      "text": "骑着一头北极熊",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 595.48,
      "end": 597.22,
      "text": "而且这实际上是我见过的",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 597.22,
      "end": 598.8,
      "text": "所有中国AI模型中",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 598.8,
      "end": 601.22,
      "text": "生成效果最好的SVG之一",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 601.22,
      "end": 602.7,
      "text": "它确实能够让许多功能",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 602.7,
      "end": 603.9,
      "text": "实现动画效果",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 603.9,
      "end": 605.86,
      "text": "所以看到他能够做到这样的事情",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 605.86,
      "end": 606.96,
      "text": "真的很酷",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 606.96,
      "end": 608.02,
      "text": "接下来看看Claude",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 608.02,
      "end": 609.72,
      "text": "Anthropic刚刚进行了一次",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 609.72,
      "end": 611.62,
      "text": "相当重大的智能体更新",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 611.62,
      "end": 613.94,
      "text": "因为Computer Use 浏览器工具",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 613.94,
      "end": 616.04,
      "text": "Skills API和Files API",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 616.04,
      "end": 618.42,
      "text": "现在已在Cloud平台上全面开放",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 618.42,
      "end": 619.68,
      "text": "这里的核心理念是",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 619.68,
      "end": 620.78,
      "text": "Cloud的智能体",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 620.78,
      "end": 622.84,
      "text": "现在不只是调用普通API了",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 622.84,
      "end": 624.1,
      "text": "借助Computer Use",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 624.1,
      "end": 626.54,
      "text": "Cloud可以与甚至没有API的应用",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 626.54,
      "end": 627.58,
      "text": "进行交互",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 627.58,
      "end": 628.36,
      "text": "而浏览器工具",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 628.36,
      "end": 630.32,
      "text": "则让它能够直接在整个网络中",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 630.32,
      "end": 631.74,
      "text": "导航和执行操作",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 631.74,
      "end": 633.14,
      "text": "然后还有Skills API",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 633.14,
      "end": 635.14,
      "text": "它允许开发者为Cloud的智能体",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 635.14,
      "end": 636.94,
      "text": "提供可复用的技能和能力",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 636.94,
      "end": 638.8,
      "text": "Files API则会让智能体",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 638.8,
      "end": 642.06,
      "text": "更轻松的在不同任务中处理同一批文档和文件",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 642.06,
      "end": 642.76,
      "text": "因此",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 642.76,
      "end": 647.18,
      "text": "Anthropic显然正致力于打造一个完整的云端托管智能体系统",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 647.18,
      "end": 649.66,
      "text": "使其能够真正完成从Computer Use",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 649.66,
      "end": 654.06,
      "text": "浏览器工具到Skills API和Files API等各种不同的操作",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 654.06,
      "end": 656.54,
      "text": "如果你喜欢这段视频并愿意支持本频道",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 656.54,
      "end": 660.24,
      "text": "可以考虑通过下方的超级感谢选项向我的频道捐赠",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 660.24,
      "end": 663.08,
      "text": "或者你也可以考虑加入我们的私人Discord服务器",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 663.08,
      "end": 663.66,
      "text": "在那里",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 663.66,
      "end": 666.76,
      "text": "你每月都可以免费使用多种AI工具的订阅服务",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 666.76,
      "end": 668.34,
      "text": "还能获取每日AI新闻",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 668.34,
      "end": 669.1,
      "text": "独家内容",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 669.1,
      "end": 670.24,
      "text": "以及更多福利",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 670.24,
      "end": 671.4,
      "text": "或许如果我有时间",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 671.4,
      "end": 673.36,
      "text": "我可能会专门制作一期完整视频",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 673.36,
      "end": 674.44,
      "text": "来介绍ONEK 1.5",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 674.44,
      "end": 676.92,
      "text": "因为这确实是一个非常出色的本地模型",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 676.92,
      "end": 678.14,
      "text": "能力相当强",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 678.14,
      "end": 679.42,
      "text": "显然它在某些方面",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 679.42,
      "end": 681.42,
      "text": "无法与Qing 3.8 Max相提并论",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 681.42,
      "end": 684.08,
      "text": "但你仍然可以从中获得很多成果",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 684.08,
      "end": 686.02,
      "text": "它在不少任务上都表现得相当不错",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 686.02,
      "end": 687.9,
      "text": "当然使用更低的位数时",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 687.9,
      "end": 689.76,
      "text": "输出质量也会有所下降",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 689.76,
      "end": 691.52,
      "text": "如果你实际运行这个模型的",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 691.52,
      "end": 693.32,
      "text": "350亿参数5Q版本",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 693.32,
      "end": 694.5,
      "text": "它能够为你带来一些",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 694.5,
      "end": 696.2,
      "text": "令人难以置信的结果",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 696.76,
      "end": 698.52,
      "text": "临时需要马上补充编码的新闻",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 698.52,
      "end": 699.16,
      "text": "据报道",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 699.16,
      "end": 701.42,
      "text": "ChatJPT Work和Codex用户",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 701.42,
      "end": 704.0,
      "text": "今天似乎即将演来一个相当不错的惊喜",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 704.0,
      "end": 705.6,
      "text": "因为Table今天宣布了一项",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 705.6,
      "end": 708.18,
      "text": "以内置的使用额度重置功能",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 708.18,
      "end": 709.58,
      "text": "而这一消息发布之际",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 709.58,
      "end": 712.74,
      "text": "Codex的活跃用户数已经突破2000万",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 712.74,
      "end": 714.62,
      "text": "这确实相当惊人",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 714.62,
      "end": 715.12,
      "text": "本质上",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 715.12,
      "end": 718.1,
      "text": "这将为用户提供一批全新的Codex使用额度",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 718.1,
      "end": 719.12,
      "text": "供他们消耗",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 719.12,
      "end": 719.74,
      "text": "因此",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 719.74,
      "end": 721.82,
      "text": "你的速率限制将恢复正常",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 721.82,
      "end": 723.86,
      "text": "或者说恢复到100%",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 723.86,
      "end": 724.42,
      "text": "这意味着",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 724.42,
      "end": 726.94,
      "text": "你可以在整个周末继续使用Codex",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 726.94,
      "end": 728.98,
      "text": "不过今天的视频就到这里了",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 728.98,
      "end": 731.18,
      "text": "我会把今天视频中使用的所有链接",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 731.18,
      "end": 732.74,
      "text": "放在下方的描述栏里",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 732.74,
      "end": 735.7,
      "text": "记得去看看World of AI Insider Club School社区",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 735.7,
      "end": 737.0,
      "text": "记得订阅第二频道",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 737.0,
      "end": 738.16,
      "text": "加入新闻通讯",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 738.16,
      "end": 739.02,
      "text": "加入Discord",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 739.02,
      "end": 740.38,
      "text": "在Twitter上关注我",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 740.38,
      "end": 741.9,
      "text": "最后记得订阅频道",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 741.9,
      "end": 742.72,
      "text": "打开通知铃",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 742.72,
      "end": 743.96,
      "text": "给这个视频点赞",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 743.96,
      "end": 745.62,
      "text": "也请看看我们之前的视频",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 745.62,
      "end": 748.06,
      "text": "以便及时了解最新的AI新闻",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 748.06,
      "end": 748.56,
      "text": "说到这里",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 748.56,
      "end": 749.14,
      "text": "朋友们",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 749.14,
      "end": 750.6,
      "text": "祝大家度过美好的一天",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 750.6,
      "end": 751.82,
      "text": "传播积极正能量",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 751.82,
      "end": 752.84,
      "text": "我们很快再见",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    },
    {
      "start": 752.84,
      "end": 753.7,
      "text": "愿他受苦",
      "language": "zh",
      "chunk": 1,
      "local_language_run": 0,
      "language_run": 0
    }
  ],
  "segmented_multilingual_transcription": {
    "status": "STABLE_SEGMENTED_MULTILINGUAL_OK",
    "created_at": "2026-08-24T12:42:43",
    "manifest": "/Users/sagawa/AI_WORK/video_notes/url/20260824_123233__中配_Gemini_4.0大量泄露_DeepSeek_V4_Flash_Vision与V5爆料_Ox_Alpha竟是MiniMax_及更多_AI资讯/segmented_work/chunk_manifest.json",
    "audio": "/Users/sagawa/AI_WORK/video_notes/url/20260824_123233__中配_Gemini_4.0大量泄露_DeepSeek_V4_Flash_Vision与V5爆料_Ox_Alpha竟是MiniMax_及更多_AI资讯/audio_16k_mono.wav",
    "audio_sha256": "f4f0ff65e32a42f4dd9d4706cd0093e768f431fa85db3c208c4eb358c45756aa",
    "audio_duration": 754.112,
    "model": "/Users/sagawa/AI_WORK/cache/huggingface/hub/models--mlx-community--whisper-large-v3-turbo/snapshots/a4aaeec0636e6fef84abdcbe3544cb2bf7e9f6fb",
    "language_profile": "/Users/sagawa/AI_WORK/video_notes/url/20260824_123233__中配_Gemini_4.0大量泄露_DeepSeek_V4_Flash_Vision与V5爆料_Ox_Alpha竟是MiniMax_及更多_AI资讯/multilingual_work/multilingual_language_profile.json",
    "dominant_languages": [
      "zh"
    ],
    "suppressed_languages": [],
    "chunk_count": 2,
    "completed_chunks": 2,
    "resumed_chunks": 0,
    "merged_segment_count": 390,
    "detected_languages": [
      "zh"
    ],
    "transcript_languages": [
      "zh"
    ],
    "missing_languages": [],
    "unexpected_languages": [],
    "language_sequence": [
      "zh"
    ],
    "speech_region_seconds": 754.112,
    "speech_covered_seconds": 750.2800000000001,
    "speech_coverage_ratio": 0.9949185266909957,
    "tail_repeat_run": 1,
    "tail_repeat_text": "AI新闻不断涌现",
    "chunk_needs_review": [],
    "qc_ok": true,
    "qc_errors": [],
    "complete_review_artifact": true
  }
}