[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fC4mfOmvgFggKrVjF80RWBc40lzUwzEwMwYDDRUZCZIE":3,"$fiLs5NNvJvA7VS7XMypEv1EQrMxq0xIZyV5ttoDTuYYI":223},[4,29,39,50,61,71,80,93,103,117,130,140,150,160,170,179,187,194,204,214],{"id":5,"title":6,"cover":7,"intro":8,"author":9,"categoryId":10,"categoryName":11,"click":12,"sort":12,"createTime":13,"tags":14},21,"Grok Bot、Muse、Dots、Cue 全方位对比（2026年11月）：AI 队长、个人助手、常驻执行者与流程自动化怎么选","\u002Farticles\u002Fart-11.png","一张流传很广的对比图把四款常驻智能体放在了同一张表里。本文按它的框架——产品定位、核心能力、记忆上下文、工具接入、自动化程度、协作方式——逐项核实并拆解，给出四类人群的选型答案与一句话结论。","AI导航站评测组",1011,"AI评测",0,"2026-11-02 12:45:36",[15,20,24],{"id":16,"name":17,"color":18,"type":19},13,"横评对比","#C2185B",1,{"id":21,"name":22,"color":23,"type":19},12,"深度评测","#00838F",{"id":25,"name":26,"color":27,"type":28},16,"智能体","#10b981",2,{"id":30,"title":31,"cover":32,"intro":33,"author":9,"categoryId":10,"categoryName":11,"click":12,"sort":12,"createTime":34,"tags":35},20,"常驻智能体大战（2026年10月）：OpenAI Dots、Meta Muse、Claude Duet、Manus Cue 四款怎么选","\u002Farticles\u002Fart-05.png","OpenAI Dots、Meta Muse、Claude Duet、Manus Cue 在 26 天内先后登场，把 AI 从「柜台答问」推进到「工位常驻」。本文用运行形态、生态入口、信任边界、可用门槛四把尺子横评四款，并给出分人群选型结论。","2026-10-30 10:48:43",[36,37,38],{"id":16,"name":17,"color":18,"type":19},{"id":21,"name":22,"color":23,"type":19},{"id":25,"name":26,"color":27,"type":28},{"id":40,"title":41,"cover":42,"intro":43,"author":9,"categoryId":10,"categoryName":11,"click":12,"sort":44,"createTime":45,"tags":46},19,"国内 Vibe Coding 生态横评（2026年10月）：Qoder、Trae、CodeBuddy、通义灵码、秒哒、吐司六款怎么选","\u002Farticles\u002Fart-10.png","国内 Vibe Coding 一年冒出 15 款以上产品，字节、阿里、腾讯各抢一端。本文按用户分层拆解 Qoder、Trae、CodeBuddy、通义灵码、秒哒、吐司六款，看它们各自在解哪一道题。",99,"2026-10-24 21:25:23",[47,48,49],{"id":16,"name":17,"color":18,"type":19},{"id":21,"name":22,"color":23,"type":19},{"id":25,"name":26,"color":27,"type":28},{"id":51,"title":52,"cover":53,"intro":54,"author":9,"categoryId":10,"categoryName":11,"click":12,"sort":55,"createTime":56,"tags":57},18,"AI 漫剧工具横评（2026年10月）：纳米、万兴剧厂、WorkRally、阅文 OPC、天工、Seko 六款怎么选","\u002Farticles\u002Fart-02.png","220 亿市场、0.47% 爆款率——2026 年的 AI 漫剧，拼的早已不是能不能生成，而是能不能稳定产出精品。本文横评纳米大片流水线、万兴剧厂、腾讯 WorkRally、阅文 OPC、天工短剧工作台与商汤 Seko 六款主流工具，讲清各自最擅长的那一环，并给出按团队类型的选型建议。",98,"2026-10-22 11:53:07",[58,59,60],{"id":16,"name":17,"color":18,"type":19},{"id":21,"name":22,"color":23,"type":19},{"id":25,"name":26,"color":27,"type":28},{"id":62,"title":63,"cover":64,"intro":65,"author":9,"categoryId":10,"categoryName":11,"click":12,"sort":12,"createTime":66,"tags":67},17,"Harness 与 Agent 到底差在哪？2026 秋季深度评测（附国产工具对照）","\u002Farticles\u002Fart-14.png","同一个模型换个 Harness，Terminal Bench 得分从 52.8% 跳到 66.5%；而多智能体团队花 1.8 到 5.1 倍成本，质量却几乎测不出差别。2026 年决定 AI 工具好不好用的，早已不是模型本身。本文拆清 Harness 与 Agent 的分层边界，并对照国产工具给出选型建议。","2026-10-15 08:47:21",[68,69,70],{"id":21,"name":22,"color":23,"type":19},{"id":16,"name":17,"color":18,"type":19},{"id":25,"name":26,"color":27,"type":28},{"id":25,"title":72,"cover":73,"intro":74,"author":9,"categoryId":10,"categoryName":11,"click":12,"sort":12,"createTime":75,"tags":76},"企业 AI 智能体横评（2026年10月）：Gemini Agent、豆包工作、WorkBuddy、千问办公谁更该上？","\u002Farticles\u002Fart-13.png","谷歌云 Gemini Agent 亮相、豆包工作同期迭代，企业智能体赛道十月骤然拥挤。本文用「能不能干活、贵不贵、管得住吗」三把尺子横评四款主流产品，附上已公开定价与分场景选型结论。","2026-10-12 14:43:56",[77,78,79],{"id":16,"name":17,"color":18,"type":19},{"id":21,"name":22,"color":23,"type":19},{"id":25,"name":26,"color":27,"type":28},{"id":16,"title":81,"cover":82,"intro":83,"author":9,"categoryId":10,"categoryName":11,"click":84,"sort":85,"createTime":86,"tags":87},"GPT-6 Astra 首发实测：贵 2.5 倍的「新一代智能」，这笔钱该不该花","\u002Farticles\u002Fart-03.png","OpenAI 把新旗舰命名为「星辰」。1.05M 上下文、ARC-AGI-3 99.9%，但 API 价格贵 2.5 倍——我们用真实任务算了笔账。",3521,96,"2026-10-01 10:48:56",[88,92],{"id":89,"name":90,"color":91,"type":19},14,"新模型速览","#558B2F",{"id":21,"name":22,"color":23,"type":19},{"id":89,"title":94,"cover":95,"intro":96,"author":9,"categoryId":10,"categoryName":11,"click":97,"sort":98,"createTime":99,"tags":100},"上线 12 天下载量反超 ChatGPT：Meta Muse 到底做对了什么","\u002Farticles\u002Fart-08.png","5 天 73 万、13 天全网 250 万+，登顶美区免费榜。Muse 爆火的答案不在「模型有多强」，而在它把 AI 从「给答案」变成了「做事情」。",4102,94,"2026-09-30 10:48:56",[101,102],{"id":25,"name":26,"color":27,"type":28},{"id":21,"name":22,"color":23,"type":19},{"id":104,"title":105,"cover":32,"intro":106,"author":9,"categoryId":10,"categoryName":11,"click":107,"sort":108,"createTime":109,"tags":110},8,"OpenClaw 深度体验三十天：开源「龙虾」能否承包你的日常任务","连续 30 天把 OpenClaw 当数字员工用：任务完成率、误操作、失控成本，以及普通人的正确打开方式。",1823,78,"2026-09-29 20:58:41",[111,115,116],{"id":112,"name":113,"color":114,"type":19},4,"开源","#909399",{"id":21,"name":22,"color":23,"type":19},{"id":25,"name":26,"color":27,"type":28},{"id":118,"title":119,"cover":120,"intro":121,"author":9,"categoryId":10,"categoryName":11,"click":122,"sort":123,"createTime":124,"tags":125},15,"苹果终于交卷：全新 Siri AI 实测，会中文了，但首发没大陆的份","\u002Farticles\u002Fart-09.png","9 月 10 日凌晨，依托 Apple Intelligence 重构的 Siri AI 随秋季发布会亮相：语义理解、多轮对话、16 种语言支持简繁中文——然后，中国大陆不在首发名单里。",2980,92,"2026-09-29 10:48:56",[126,129],{"id":62,"name":127,"color":128,"type":28},"AI硬件","#f59e0b",{"id":16,"name":17,"color":18,"type":19},{"id":131,"title":132,"cover":133,"intro":134,"author":9,"categoryId":10,"categoryName":11,"click":135,"sort":136,"createTime":137,"tags":138},9,"2026 AI 编程工具终局之战：Claude Code、Qoder、CodeBuddy 谁在真提效","\u002Farticles\u002Fart-15.png","同一个真实项目交给四款工具各做一遍：一次通过率、记忆系统成色、夜间无人值守表现，数据说话。",3120,76,"2026-09-27 20:58:41",[139],{"id":16,"name":17,"color":18,"type":19},{"id":141,"title":142,"cover":7,"intro":143,"author":9,"categoryId":10,"categoryName":11,"click":144,"sort":145,"createTime":146,"tags":147},10,"AI 眼镜怎么选：雷鸟 iO、讯飞、ROKID 三款横评","上半年 70 款新品井喷后，我们挑了三款代表机型佩戴两周：34g 光波导、48 小时续航、超级助理，谁才是第一副「可穿戴大脑」。",2687,74,"2026-09-25 20:58:41",[148,149],{"id":16,"name":17,"color":18,"type":19},{"id":62,"name":127,"color":128,"type":28},{"id":151,"title":152,"cover":64,"intro":153,"author":9,"categoryId":10,"categoryName":11,"click":154,"sort":155,"createTime":156,"tags":157},11,"企业级 Agent 落地三个月实录：为什么 88% 的公司在投入，只有 24% 赚到钱","毕马威说 88% 的组织在投资 Agent，Gartner 说只有 17% 完成部署。我们跟踪了三家真实企业的落地全程，拆出成功与翻车的分界线。",2411,72,"2026-09-23 20:58:41",[158,159],{"id":21,"name":22,"color":23,"type":19},{"id":25,"name":26,"color":27,"type":28},{"id":21,"title":161,"cover":73,"intro":162,"author":9,"categoryId":10,"categoryName":11,"click":163,"sort":164,"createTime":165,"tags":166},"MCP 生态半年观察：AI 的「USB 接口」正在连接一切","MCP、A2A 协议标准化半年后：工具调用成功率翻了多少？哪些场景已即插即用？开发者红利期还有多久？",1567,70,"2026-09-21 20:58:41",[167,168,169],{"id":112,"name":113,"color":114,"type":19},{"id":21,"name":22,"color":23,"type":19},{"id":25,"name":26,"color":27,"type":28},{"id":28,"title":171,"cover":172,"intro":173,"author":9,"categoryId":10,"categoryName":11,"click":174,"sort":175,"createTime":176,"tags":177},"DeepSeek-R1 深度评测：国产推理之王的边界在哪里","\u002Farticles\u002Fart-06.png","我们用 40 道真题、8 类真实任务实测 DeepSeek-R1：数学与代码接近满分的背后，普通用户真正该关心的优缺点都在这里。",3241,100,"2026-09-18 16:41:14",[178],{"id":21,"name":22,"color":23,"type":19},{"id":180,"title":181,"cover":53,"intro":182,"author":9,"categoryId":10,"categoryName":11,"click":183,"sort":85,"createTime":184,"tags":185},3,"Kimi、通义千问、豆包长文本横评：谁才是「读论文神器」","同一篇 96 页 PDF 财报，三个模型轮流读：定位准确率、幻觉率、交叉引用能力，差距比想象中大。",2876,"2026-09-16 16:41:14",[186],{"id":16,"name":17,"color":18,"type":19},{"id":112,"title":188,"cover":189,"intro":190,"author":9,"categoryId":10,"categoryName":11,"click":97,"sort":123,"createTime":191,"tags":192},"可灵AI 视频生成实测：距离「AI导演」还有多远","\u002Farticles\u002Fart-07.png","用可灵生成了 30 段视频：物理效果惊艳，但人物一致性仍是硬伤。附 6 条实用提示词技巧。","2026-09-13 16:41:14",[193],{"id":89,"name":90,"color":91,"type":19},{"id":195,"title":196,"cover":197,"intro":198,"author":9,"categoryId":10,"categoryName":11,"click":199,"sort":200,"createTime":201,"tags":202},5,"秘塔AI搜索三个月使用报告：没有广告的搜索是什么体验","\u002Farticles\u002Fart-04.png","连续 90 天把秘塔设为默认搜索引擎：信息溯源、学术模式、三档检索深度怎么选，一次说清。",1988,88,"2026-09-09 16:41:14",[203],{"id":89,"name":90,"color":91,"type":19},{"id":205,"title":206,"cover":207,"intro":208,"author":9,"categoryId":10,"categoryName":11,"click":209,"sort":210,"createTime":211,"tags":212},6,"2026 国产大模型编程横评：DeepSeek、GLM、通义写同一个项目谁最快","\u002Farticles\u002Fart-12.png","同一个全栈待办应用，三个模型各写一遍：一次通过率、代码风格、返工成本，数据说话。",3567,84,"2026-09-05 16:41:14",[213],{"id":16,"name":17,"color":18,"type":19},{"id":215,"title":216,"cover":42,"intro":217,"author":9,"categoryId":10,"categoryName":11,"click":218,"sort":219,"createTime":220,"tags":221},7,"剪映AI「图文成片」实测：零基础 30 分钟做出一条像样的短视频","从文案到成片全流程记录：AI 配音、智能字幕、素材匹配的真实水平，以及三个必须手动救的坑。",2245,80,"2026-09-01 16:41:14",[222],{"id":89,"name":90,"color":91,"type":19},[224,225,226,227,228,229],{"id":21,"name":22,"color":23,"type":19,"toolCount":21},{"id":16,"name":17,"color":18,"type":19,"toolCount":151},{"id":89,"name":90,"color":91,"type":19,"toolCount":112},{"id":112,"name":113,"color":114,"type":19,"toolCount":28},{"id":25,"name":26,"color":27,"type":28,"toolCount":141},{"id":62,"name":127,"color":128,"type":28,"toolCount":28}]