[{"data":1,"prerenderedAt":9673},["ShallowReactive",2],{"header-counts":3,"footer-counts":6,"tool-\u002Ftools\u002Fagent\u002Fplatform\u002Fflowise":8,"cat-rank-agent-platform":620,"tool-related-agent\u002Fplatform\u002Fflowise":7915,"tool-reviews-agent\u002Fplatform\u002Fflowise":7916,"tool-alts-agent\u002Fplatform\u002Fflowise":7917},{"tools":4,"reviews":5},98,27,{"tools":4,"reviews":5,"playbooks":7,"news":5},22,{"id":9,"title":10,"alternatives":11,"api_compatible":15,"body":16,"category":584,"chinese_friendly":570,"cover":585,"description":586,"domestic":587,"extension":588,"faq":15,"free":587,"github":565,"languages":589,"lastVerified":591,"meta":592,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":594,"pillar":595,"platforms":596,"priceTable":15,"pricing":599,"published":600,"relatedPlaybooks":15,"relatedReviews":15,"score":601,"self_host":587,"seo":604,"seoTitle":605,"slug":606,"sources":607,"stem":610,"suitable":15,"tagline":611,"tags":612,"updated":591,"verdict":618,"website":557,"__hash__":619},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fflowise.md","Flowise",[12,13,14],"agent\u002Fplatform\u002Flangflow","agent\u002Fplatform\u002Fdify","agent\u002Fplatform\u002Fn8n",null,{"type":17,"value":18,"toc":568},"minimark",[19,24,28,31,34,93,96,162,168,172,180,205,210,233,236,265,268,409,412,456,460,492,496,502,508,518,524,527,543,546,551],[20,21,23],"h2",{"id":22},"tldr","TL;DR",[25,26,27],"p",{},"Flowise 是开源 LLM 流程编排平台（Apache 2.0），拖拽式可视化构建 AI 应用，底层基于 LangChain \u002F LlamaIndex 生态。节点包括 LLM \u002F Chat Model \u002F Embedding \u002F Vector Store \u002F Tool \u002F Agent \u002F Memory 等，连线编排成完整流程。Docker 自托管 + Cloud 云端双模式，导出为 API \u002F 嵌入式聊天组件 \u002F SDK。",[25,29,30],{},"适合：不写代码也要搭建 AI 应用的产品\u002F运营团队、快速验证 RAG \u002F Chatbot 原型、LangChain 生态用户可视化探索。不适合：复杂业务逻辑编排（用 n8n \u002F Dify）、大规模并发生产服务、深度定制需求（直接写 LangChain 代码）。",[20,32,33],{"id":33},"核心能力",[35,36,37,45,51,57,63,69,75,81,87],"ul",{},[38,39,40,44],"li",{},[41,42,43],"strong",{},"可视化拖拽编排","：节点 + 连线构建 LLM 流程，实时预览",[38,46,47,50],{},[41,48,49],{},"LangChain 生态","：直接使用 LangChain \u002F LlamaIndex 全部组件",[38,52,53,56],{},[41,54,55],{},"丰富节点","：LLM \u002F Chat Model \u002F Embedding \u002F Vector Store \u002F Tool \u002F Agent \u002F Memory \u002F Chain",[38,58,59,62],{},[41,60,61],{},"Agent 支持","：Conversational Agent \u002F Tool Calling Agent \u002F ReAct Agent",[38,64,65,68],{},[41,66,67],{},"RAG 流程","：文档加载 → 切片 → 嵌入 → 向量存储 → 检索 → 生成，全可视化",[38,70,71,74],{},[41,72,73],{},"多向量数据库","：Pinecone \u002F Qdrant \u002F Chroma \u002F Weaviate \u002F Supabase",[38,76,77,80],{},[41,78,79],{},"多模型接入","：OpenAI \u002F Claude \u002F Gemini \u002F Azure \u002F Ollama \u002F HuggingFace",[38,82,83,86],{},[41,84,85],{},"部署方式","：导出 REST API \u002F 嵌入式聊天组件 \u002F React SDK",[38,88,89,92],{},[41,90,91],{},"凭据管理","：API Key 加密存储，支持环境变量",[20,94,95],{"id":95},"价格",[97,98,99,114],"table",{},[100,101,102],"thead",{},[103,104,105,109,111],"tr",{},[106,107,108],"th",{},"方案",[106,110,95],{},[106,112,113],{},"核心功能",[115,116,117,129,140,151],"tbody",{},[103,118,119,123,126],{},[120,121,122],"td",{},"开源版",[120,124,125],{},"$0",[120,127,128],{},"完整功能，Apache 2.0，自托管",[103,130,131,134,137],{},[120,132,133],{},"Cloud Starter",[120,135,136],{},"$39\u002F月起",[120,138,139],{},"托管服务，1 个工作区",[103,141,142,145,148],{},[120,143,144],{},"Cloud Pro",[120,146,147],{},"$89\u002F月起",[120,149,150],{},"多工作区 + 团队协作 + 高并发",[103,152,153,156,159],{},[120,154,155],{},"Enterprise",[120,157,158],{},"联系销售",[120,160,161],{},"私有部署 + SSO + 专属支持",[163,164,165],"blockquote",{},[25,166,167],{},"价格信息基于 2026-07 官网，可能调整。",[20,169,171],{"id":170},"体验与评测资料整理","体验与评测（资料整理）",[163,173,174],{},[25,175,176,177],{},"说明：本节基于官方文档与公开评测整理，非本站独立实测环境，具体数据请以官方为准。\n",[41,178,179],{},"亮点：",[35,181,182,185,188,191,199,202],{},[38,183,184],{},"拖拽编排体验流畅，LangChain 组件全覆盖，不用写代码也能搭复杂流程",[38,186,187],{},"RAG 流程模板开箱即用，上传 PDF + 接 OpenAI 几分钟出问答机器人",[38,189,190],{},"嵌入式聊天组件方便，生成一段 JS 代码嵌入网页即可",[38,192,193,194,198],{},"Docker 部署简单，",[195,196,197],"code",{},"docker-compose up"," 一条命令",[38,200,201],{},"节点参数可视化配置，temperature \u002F chunk size 等滑块调整直观",[38,203,204],{},"社区活跃，模板市场有不少现成的 Chatflow 可复用",[25,206,207],{},[41,208,209],{},"踩坑：",[35,211,212,215,218,221,224,227,230],{},[38,213,214],{},"流程复杂后画布混乱，连线交叉难以维护",[38,216,217],{},"性能一般——每个请求经过多个节点序列化\u002F反序列化，延迟偏高",[38,219,220],{},"错误信息不够友好，节点报错时定位问题费时",[38,222,223],{},"升级 LangChain 版本后部分节点可能不兼容",[38,225,226],{},"无法版本控制流程结构，团队协作时容易覆盖",[38,228,229],{},"并发能力有限，高并发场景需配合队列 + 负载均衡",[38,231,232],{},"深度定制仍需写代码——自定义节点门槛不低",[20,234,235],{"id":235},"上手",[237,238,239,246,253,256,259,262],"ol",{},[38,240,241,242,245],{},"Docker 部署：",[195,243,244],{},"docker-compose up -d","（官方提供 docker-compose.yml）",[38,247,248,249,252],{},"访问 ",[195,250,251],{},"http:\u002F\u002Flocalhost:3000","，创建管理员账号",[38,254,255],{},"配置 Credential：添加 OpenAI API Key 等凭据",[38,257,258],{},"新建 Chatflow → 从空白或模板开始",[38,260,261],{},"拖入节点：Chat OpenAI + Conversational Retrieval Chain + Vector Store",[38,263,264],{},"连线编排 → 右上角 Save → 测试对话 → 导出 API \u002F 嵌入组件",[20,266,267],{"id":267},"对比",[97,269,270,288],{},[100,271,272],{},[103,273,274,277,279,282,285],{},[106,275,276],{},"维度",[106,278,10],{},[106,280,281],{},"Langflow",[106,283,284],{},"Dify",[106,286,287],{},"n8n",[115,289,290,307,323,337,354,367,380,394],{},[103,291,292,295,298,301,304],{},[120,293,294],{},"编排方式",[120,296,297],{},"拖拽 Chatflow",[120,299,300],{},"拖拽 Flow",[120,302,303],{},"拖拽 + YAML",[120,305,306],{},"拖拽 Workflow",[103,308,309,312,315,317,320],{},[120,310,311],{},"生态",[120,313,314],{},"LangChain",[120,316,314],{},[120,318,319],{},"自有",[120,321,322],{},"通用自动化",[103,324,325,327,330,332,334],{},[120,326,235],{},[120,328,329],{},"低",[120,331,329],{},[120,333,329],{},[120,335,336],{},"中",[103,338,339,342,345,348,351],{},[120,340,341],{},"RAG",[120,343,344],{},"✅ 模板丰富",[120,346,347],{},"✅",[120,349,350],{},"✅ 强",[120,352,353],{},"需自建",[103,355,356,359,361,363,365],{},[120,357,358],{},"Agent",[120,360,347],{},[120,362,347],{},[120,364,350],{},[120,366,347],{},[103,368,369,372,374,376,378],{},[120,370,371],{},"API 导出",[120,373,347],{},[120,375,347],{},[120,377,347],{},[120,379,347],{},[103,381,382,385,388,390,392],{},[120,383,384],{},"部署",[120,386,387],{},"Docker",[120,389,387],{},[120,391,387],{},[120,393,387],{},[103,395,396,399,402,404,407],{},[120,397,398],{},"适合",[120,400,401],{},"AI 应用原型",[120,403,401],{},[120,405,406],{},"生产 AI 应用",[120,408,322],{},[20,410,411],{"id":411},"避坑",[35,413,414,420,426,432,438,444,450],{},[38,415,416,419],{},[41,417,418],{},"流程别太复杂","：超过 15 个节点的 Chatflow 维护成本急升，拆分成多个",[38,421,422,425],{},[41,423,424],{},"性能优化","：合并可合并的节点，减少序列化开销",[38,427,428,431],{},[41,429,430],{},"版本管理","：定期导出 Chatflow JSON 备份，升级前测兼容性",[38,433,434,437],{},[41,435,436],{},"凭据安全","：不要在流程中硬编码 API Key，统一走 Credential 管理",[38,439,440,443],{},[41,441,442],{},"并发测试","：上线前压测，Flowise 单实例并发有限",[38,445,446,449],{},[41,447,448],{},"LangChain 版本","：关注 Flowise 更新日志，LangChain 大版本升级可能有 breaking change",[38,451,452,455],{},[41,453,454],{},"别替代生产框架","：复杂生产应用还是用 Dify 或直接写代码",[20,457,459],{"id":458},"适合-不适合","适合 \u002F 不适合",[35,461,462,465,468,471,474,477,480,483,486,489],{},[38,463,464],{},"✅ 不写代码搭建 RAG \u002F Chatbot 原型",[38,466,467],{},"✅ LangChain 生态用户可视化探索",[38,469,470],{},"✅ 快速验证 AI 应用概念",[38,472,473],{},"✅ 嵌入式聊天组件场景",[38,475,476],{},"✅ 教学 \u002F 演示 LLM 流程",[38,478,479],{},"❌ 复杂业务逻辑编排（用 n8n \u002F Dify）",[38,481,482],{},"❌ 大规模并发生产服务（性能有限）",[38,484,485],{},"❌ 深度定制需求（直接写 LangChain 代码）",[38,487,488],{},"❌ 需要版本控制 + 团队协作开发（能力有限）",[38,490,491],{},"❌ 非 LangChain 生态需求",[20,493,495],{"id":494},"faq","FAQ",[25,497,498,501],{},[41,499,500],{},"Q: Flowise 和 Langflow 怎么选？","\nA: 两者定位几乎相同——都是 LangChain 可视化编排。Flowise 界面更简洁、上手稍快、社区模板多。Langflow 由 DataStax 维护、与 LangChain 官方关系更近、组件更新更快。都试试选顺手的即可。",[25,503,504,507],{},[41,505,506],{},"Q: Flowise 和 Dify 怎么选？","\nA: Flowise 专注 LLM 流程可视化编排，轻量、原型验证快。Dify 是完整 AI 应用开发平台——工作流 + Agent + RAG + API 管理 + 监控，功能更全更适合生产。做原型选 Flowise，做产品选 Dify。",[25,509,510,513,514,517],{},[41,511,512],{},"Q: 可以接入本地模型吗？","\nA: 可以。Flowise 支持 Ollama \u002F HuggingFace 本地模型节点，配置 Ollama 地址（",[195,515,516],{},"http:\u002F\u002Flocalhost:11434","）即可在流程中使用本地 LLM 和 Embedding 模型。",[25,519,520,523],{},[41,521,522],{},"Q: 生产环境能用吗？","\nA: 小规模可以（内部工具 \u002F 低并发场景）。高并发生产环境建议用 Dify 或直接写代码——Flowise 的节点序列化开销和单实例并发限制是瓶颈。如需生产部署，配合 Nginx 负载均衡 + 多实例 + Redis 队列。",[20,525,526],{"id":526},"相关阅读",[25,528,529,534,535,534,539],{},[530,531,533],"a",{"href":532},"\u002Fagent\u002Fplatform\u002Fautogen.html","AutoGen"," · ",[530,536,538],{"href":537},"\u002Fagent\u002Fplatform\u002Fcrewai.html","CrewAI",[530,540,542],{"href":541},"\u002Fagent\u002Fplatform\u002Fragflow.html","RAGFlow",[20,544,545],{"id":545},"来源",[163,547,548],{},[25,549,550],{},"本文的价格、版本号与性能数据均参考以下官方渠道整理，可能随时间变动，请以官方实时信息为准。",[35,552,553,561],{},[38,554,555],{},[530,556,560],{"href":557,"rel":558},"https:\u002F\u002Fflowiseai.com",[559],"nofollow","官网",[38,562,563],{},[530,564,567],{"href":565,"rel":566},"https:\u002F\u002Fgithub.com\u002FFlowiseAI\u002FFlowise",[559],"GitHub",{"title":569,"searchDepth":570,"depth":570,"links":571},"",3,[572,574,575,576,577,578,579,580,581,582,583],{"id":22,"depth":573,"text":23},2,{"id":33,"depth":573,"text":33},{"id":95,"depth":573,"text":95},{"id":170,"depth":573,"text":171},{"id":235,"depth":573,"text":235},{"id":267,"depth":573,"text":267},{"id":411,"depth":573,"text":411},{"id":458,"depth":573,"text":459},{"id":494,"depth":573,"text":495},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},"platform","\u002Fimg\u002Ftools\u002Fflowise.webp","Flowise 真实评测：开源 LLM 流程编排平台（Apache 2.0 协议），拖拽式可视化构建 AI 应用，基于 LangChain 生态。支持 Docker 自托管 + Cloud 云端，适合不写代码也能搭建 RAG\u002FAgent\u002FChatbot 流程的团队。",false,"md",[590],"en","2026-07-30",{},true,"\u002Ftools\u002Fagent\u002Fplatform\u002Fflowise","agent",[597,598],"linux","docker","Free \u002F 开源（Apache 2.0）\u002F Cloud","2026-07-05",{"power":570,"ux":602,"price":603,"cn_support":570,"stability":570},4,5,{"title":10,"description":586},"Flowise - 拖拽式 LLM 流程编排评测 | AIHO","agent\u002Fplatform\u002Fflowise",[608,609],{"title":560,"url":557},{"title":567,"url":565},"tools\u002Fagent\u002Fplatform\u002Fflowise","开源 LLM 流程编排，拖拽式构建 AI 应用",[613,614,615,616,617],"agent-platform","opensource","flow","no-code","langchain","不写代码也要拖拽搭建 RAG \u002F Chatbot \u002F Agent 流程的团队首选，基于 LangChain 生态组件丰富、可视化编排直观，但复杂流程维护难、性能一般、深度定制仍需写代码。","T4ETS7kNZnBNcywZPndJ_ynLVGjoghi1D0jMxZ7sK1A",[621,1116,1614,2523,3022,4061,5135,5543,6164,6868,7370],{"id":622,"title":623,"alternatives":624,"api_compatible":15,"body":626,"category":584,"chinese_friendly":570,"cover":1092,"description":1093,"domestic":587,"extension":588,"faq":15,"free":587,"github":1077,"languages":1094,"lastVerified":591,"meta":1095,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":1096,"pillar":595,"platforms":1097,"priceTable":15,"pricing":1100,"published":600,"relatedPlaybooks":15,"relatedReviews":15,"score":1101,"self_host":587,"seo":1102,"seoTitle":1103,"slug":1104,"sources":1105,"stem":1108,"suitable":15,"tagline":1109,"tags":1110,"updated":591,"verdict":1114,"website":1071,"__hash__":1115},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fanythingllm.md","AnythingLLM",[13,625,12],"agent\u002Fplatform\u002Ffastgpt",{"type":17,"value":627,"toc":1079},[628,630,633,636,638,693,695,738,742,744,750,773,777,797,799,823,825,946,948,989,991,1020,1022,1028,1034,1040,1046,1048,1059,1061,1065],[20,629,23],{"id":22},[25,631,632],{},"AnythingLLM 是 Mintplex Labs 出品的开源私有部署 LLM 平台，MIT 协议，主打\"一站式 RAG 知识库 + Agent + 多用户权限管理\"。桌面应用 \u002F Docker 双部署模式，接入 OpenAI \u002F Claude \u002F Ollama \u002F Azure 等任意模型，Workspaces 隔离不同知识库，内置向量数据库。适合需要私有化部署 AI 知识库且不写代码的团队。",[25,634,635],{},"适合：企业内网知识库、团队共享 AI 助手、需要多用户权限控制的私有化场景。不适合：需要复杂 Agent 编排（用 Dify \u002F Langflow）、需要极致 RAG 召回精度（用 RAGFlow \u002F FastGPT）、需要大规模并发生产服务。",[20,637,33],{"id":33},[35,639,640,646,652,658,664,669,675,681,687],{},[38,641,642,645],{},[41,643,644],{},"私有化部署","：Docker \u002F 桌面应用（Win\u002FMac\u002FLinux），数据完全在内网",[38,647,648,651],{},[41,649,650],{},"Workspaces 知识库隔离","：不同工作区独立向量库 + 文档 + 对话历史",[38,653,654,657],{},[41,655,656],{},"多用户权限管理","：管理员 \u002F 用户 \u002F 多工作区角色分配，适合团队使用",[38,659,660,663],{},[41,661,662],{},"任意模型接入","：OpenAI \u002F Claude \u002F Azure \u002F Ollama \u002F LM Studio \u002F 本地模型",[38,665,666,668],{},[41,667,73],{},"：内置 LanceDB，可选 Chroma \u002F Pinecone \u002F Weaviate \u002F Qdrant",[38,670,671,674],{},[41,672,673],{},"文档处理","：PDF \u002F Word \u002F Excel \u002F TXT \u002F Markdown \u002F 网页链接，自动切片 + 向量化",[38,676,677,680],{},[41,678,679],{},"Agent 能力","：内置 Web 搜索 \u002F RAG 搜索 \u002F SQL 查询等工具调用",[38,682,683,686],{},[41,684,685],{},"嵌入向量","：支持自定义 embedding 模型，兼容 OpenAI \u002F 本地嵌入",[38,688,689,692],{},[41,690,691],{},"API 接口","：提供完整 REST API，可集成到外部系统",[20,694,95],{"id":95},[97,696,697,707],{},[100,698,699],{},[103,700,701,703,705],{},[106,702,108],{},[106,704,95],{},[106,706,113],{},[115,708,709,718,729],{},[103,710,711,713,715],{},[120,712,122],{},[120,714,125],{},[120,716,717],{},"完整功能，MIT 协议，自托管",[103,719,720,723,726],{},[120,721,722],{},"Cloud",[120,724,725],{},"$30\u002F月起",[120,727,728],{},"托管服务，免去运维，含团队协作",[103,730,731,733,735],{},[120,732,155],{},[120,734,158],{},[120,736,737],{},"SSO \u002F 审计日志 \u002F 私有部署支持",[163,739,740],{},[25,741,167],{},[20,743,171],{"id":170},[163,745,746],{},[25,747,176,748],{},[41,749,179],{},[35,751,752,758,761,764,767,770],{},[38,753,754,755,757],{},"Docker 部署极快，一条 ",[195,756,197],{}," 起来就能用",[38,759,760],{},"Workspaces 隔离设计实用，不同部门知识库互不干扰",[38,762,763],{},"接 Ollama 本地模型完全离线运行，数据不出内网",[38,765,766],{},"桌面应用适合个人用户，安装即用零配置",[38,768,769],{},"文档上传后自动切片 + 向量化，问答效果在通用场景下可接受",[38,771,772],{},"多用户权限管理是开源 RAG 平台中少有的完整实现",[25,774,775],{},[41,776,209],{},[35,778,779,782,785,788,791,794],{},[38,780,781],{},"文档切片策略偏简单（固定长度），复杂表格 \u002F 图文混排召回效果一般",[38,783,784],{},"大文件（100MB+ PDF）处理偶尔超时，需调超时参数",[38,786,787],{},"Agent 能力有限，复杂工具链编排不如 Dify",[38,789,790],{},"向量库默认 LanceDB 在数据量大时查询变慢，建议切 Qdrant \u002F Chroma",[38,792,793],{},"UI 偶有卡顿，文档列表加载慢",[38,795,796],{},"中文文档的 OCR 需要额外配置，默认对扫描件支持有限",[20,798,235],{"id":235},[237,800,801,805,811,814,817,820],{},[38,802,241,803,245],{},[195,804,244],{},[38,806,807,808,252],{},"首次访问 ",[195,809,810],{},"http:\u002F\u002Flocalhost:3001",[38,812,813],{},"Settings → LLM Provider 配置模型（OpenAI API Key 或 Ollama 地址）",[38,815,816],{},"创建 Workspace → 上传文档（PDF\u002FWord\u002FTXT）",[38,818,819],{},"等待文档向量化完成，在 Chat 中开始问答",[38,821,822],{},"Settings → Users 添加团队成员并分配工作区权限",[20,824,267],{"id":267},[97,826,827,842],{},[100,828,829],{},[103,830,831,833,835,837,840],{},[106,832,276],{},[106,834,623],{},[106,836,284],{},[106,838,839],{},"FastGPT",[106,841,281],{},[115,843,844,858,873,887,903,917,930],{},[103,845,846,849,852,854,856],{},[120,847,848],{},"部署门槛",[120,850,851],{},"极低",[120,853,336],{},[120,855,336],{},[120,857,336],{},[103,859,860,863,866,868,870],{},[120,861,862],{},"多用户权限",[120,864,865],{},"✅ 完整",[120,867,347],{},[120,869,347],{},[120,871,872],{},"❌",[103,874,875,878,880,883,885],{},[120,876,877],{},"RAG 精度",[120,879,336],{},[120,881,882],{},"高",[120,884,882],{},[120,886,336],{},[103,888,889,892,895,898,900],{},[120,890,891],{},"Agent 编排",[120,893,894],{},"基础",[120,896,897],{},"强",[120,899,336],{},[120,901,902],{},"强（可视化）",[103,904,905,908,911,913,915],{},[120,906,907],{},"模型接入",[120,909,910],{},"丰富",[120,912,910],{},[120,914,910],{},[120,916,910],{},[103,918,919,922,924,926,928],{},[120,920,921],{},"桌面应用",[120,923,347],{},[120,925,872],{},[120,927,872],{},[120,929,872],{},[103,931,932,935,938,941,944],{},[120,933,934],{},"协议",[120,936,937],{},"MIT",[120,939,940],{},"Apache 2.0",[120,942,943],{},"FastGPT Open",[120,945,937],{},[20,947,411],{"id":411},[35,949,950,956,962,968,977,983],{},[38,951,952,955],{},[41,953,954],{},"切片策略默认偏简单","：对结构化文档（表格\u002F代码）效果差，可调 chunk size",[38,957,958,961],{},[41,959,960],{},"LanceDB 大数据量变慢","：文档超过 1 万条建议切 Qdrant 或 Chroma",[38,963,964,967],{},[41,965,966],{},"大文件超时","：调整 Docker 超时配置，或拆分文档上传",[38,969,970,973,974],{},[41,971,972],{},"Ollama 连接","：Docker 内访问宿主机 Ollama 需用 ",[195,975,976],{},"host.docker.internal",[38,978,979,982],{},[41,980,981],{},"embedding 模型选择","：中文场景建议用 bge-large-zh 而非默认 OpenAI embedding",[38,984,985,988],{},[41,986,987],{},"不要当生产级 Agent 平台用","：Agent 能力是辅助，复杂编排上 Dify",[20,990,459],{"id":458},[35,992,993,996,999,1002,1005,1008,1011,1014,1017],{},[38,994,995],{},"✅ 企业内网私有化 AI 知识库",[38,997,998],{},"✅ 团队共享 AI 助手 + 多用户权限管理",[38,1000,1001],{},"✅ 接 Ollama 完全离线运行",[38,1003,1004],{},"✅ 个人桌面端快速体验 RAG",[38,1006,1007],{},"✅ 需要快速验证 RAG 概念的原型项目",[38,1009,1010],{},"❌ 需要复杂 Agent 工作流编排（用 Dify \u002F Langflow）",[38,1012,1013],{},"❌ 需要极致 RAG 召回精度（用 RAGFlow \u002F FastGPT）",[38,1015,1016],{},"❌ 大规模并发生产服务（架构未做高可用）",[38,1018,1019],{},"❌ 需要深度文档解析（复杂表格\u002F公式\u002F扫描件）",[20,1021,495],{"id":494},[25,1023,1024,1027],{},[41,1025,1026],{},"Q: AnythingLLM 和 Dify 怎么选？","\nA: AnythingLLM 更轻量，部署快、有桌面应用、多用户权限开箱即用，适合快速搭建团队知识库。Dify 功能更全面，Agent 编排、工作流、API 发布能力更强，适合需要构建复杂 AI 应用的团队。简单知识库选 AnythingLLM，复杂应用选 Dify。",[25,1029,1030,1033],{},[41,1031,1032],{},"Q: 可以完全离线使用吗？","\nA: 可以。接 Ollama 本地模型 + 用本地 embedding 模型（如 bge-large-zh）+ 内置 LanceDB 向量库，整个系统完全离线运行，数据不出内网。适合数据敏感的企业场景。",[25,1035,1036,1039],{},[41,1037,1038],{},"Q: 免费开源版有什么限制？","\nA: MIT 协议开源版功能完整，无用户数 \u002F 文档数 \u002F API 调用限制。Cloud 版和 Enterprise 版主要是托管服务和企业管理功能（SSO \u002F 审计日志），功能层面开源版已够用。",[25,1041,1042,1045],{},[41,1043,1044],{},"Q: 支持中文文档吗？","\nA: 支持，但效果取决于 embedding 模型。默认 OpenAI embedding 对中文尚可，追求精度建议切换 bge-large-zh 或 m3e 模型。OCR 扫描件需额外配置 Tesseract 或接入外部 OCR 服务。",[20,1047,526],{"id":526},[25,1049,1050,534,1052,534,1055],{},[530,1051,542],{"href":541},[530,1053,10],{"href":1054},"\u002Fagent\u002Fplatform\u002Fflowise.html",[530,1056,1058],{"href":1057},"\u002Fagent\u002Fgeneral\u002Fperplexity.html","Perplexity",[20,1060,545],{"id":545},[163,1062,1063],{},[25,1064,550],{},[35,1066,1067,1073],{},[38,1068,1069],{},[530,1070,560],{"href":1071,"rel":1072},"https:\u002F\u002Fuseanything.com",[559],[38,1074,1075],{},[530,1076,567],{"href":1077,"rel":1078},"https:\u002F\u002Fgithub.com\u002FMintplex-Labs\u002Fanything-llm",[559],{"title":569,"searchDepth":570,"depth":570,"links":1080},[1081,1082,1083,1084,1085,1086,1087,1088,1089,1090,1091],{"id":22,"depth":573,"text":23},{"id":33,"depth":573,"text":33},{"id":95,"depth":573,"text":95},{"id":170,"depth":573,"text":171},{"id":235,"depth":573,"text":235},{"id":267,"depth":573,"text":267},{"id":411,"depth":573,"text":411},{"id":458,"depth":573,"text":459},{"id":494,"depth":573,"text":495},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},"\u002Fimg\u002Ftools\u002Fanythingllm.webp","AnythingLLM 真实评测：Mintplex Labs 出品的开源私有部署 LLM 平台（MIT 协议），一站式 RAG 知识库 + Agent + 多用户权限管理。支持 Docker\u002F桌面部署，接入 OpenAI\u002FClaude\u002FOllama 等任意模型，适合企业内网私有化 AI 知识库场景。",[590],{},"\u002Ftools\u002Fagent\u002Fplatform\u002Fanythingllm",[1098,1099,597,598],"windows","macos","Free \u002F 开源（MIT）\u002F Cloud",{"power":570,"ux":602,"price":603,"cn_support":570,"stability":570},{"title":623,"description":1093},"AnythingLLM - 开源私有部署 LLM 平台评测 | AIHO","agent\u002Fplatform\u002Fanythingllm",[1106,1107],{"title":560,"url":1071},{"title":567,"url":1077},"tools\u002Fagent\u002Fplatform\u002Fanythingllm","开源私有部署 LLM 平台，一站式 RAG + Agent + 多用户",[613,614,1111,1112,1113],"self-host","rag","multi-user","需要快速搭建私有化 AI 知识库且要求多用户权限管理的企业团队首选，MIT 协议 + 桌面\u002FDocker 双模式 + 任意模型接入降低了部署门槛，但 RAG 精度和 Agent 编排能力不及 Dify\u002FFastGPT 等专业平台。","iKAMhkQImqK_QZGWLaE4i_9IjAMMpCeSwojp4L34T6A",{"id":1117,"title":533,"alternatives":1118,"api_compatible":15,"body":1120,"category":584,"chinese_friendly":573,"cover":1591,"description":1592,"domestic":587,"extension":588,"faq":15,"free":587,"github":1576,"languages":1593,"lastVerified":591,"meta":1594,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":1595,"pillar":595,"platforms":1596,"priceTable":15,"pricing":1597,"published":600,"relatedPlaybooks":15,"relatedReviews":15,"score":1598,"self_host":587,"seo":1599,"seoTitle":1600,"slug":1601,"sources":1602,"stem":1605,"suitable":15,"tagline":1606,"tags":1607,"updated":591,"verdict":1612,"website":1570,"__hash__":1613},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fautogen.md",[1119,12,13],"agent\u002Fplatform\u002Fcrewai",{"type":17,"value":1121,"toc":1578},[1122,1124,1131,1134,1136,1192,1194,1197,1199,1205,1229,1233,1256,1258,1294,1296,1422,1424,1472,1474,1506,1508,1514,1532,1538,1546,1548,1558,1560,1564],[20,1123,23],{"id":22},[25,1125,1126,1127,1130],{},"AutoGen 是微软研究院开源的多 Agent 对话框架（MIT 协议），用 Python 代码定义 Agent 角色、对话流程和工具调用。核心是 ",[195,1128,1129],{},"ConversableAgent"," + Group Chat 模式——多个 Agent 自动对话协作完成任务，支持代码执行、工具调用、人在回路（human-in-the-loop）。AutoGen 0.4+ 重构为事件驱动架构，性能和扩展性大幅提升。",[25,1132,1133],{},"适合：AI 研究者、需要精细控制 Agent 协作逻辑的高级开发者、多 Agent 实验项目。不适合：快速原型验证（用 CrewAI）、非技术用户（用 Dify \u002F Flowise）、需要 GUI 的场景、追求 API 稳定性的生产项目。",[20,1135,33],{"id":33},[35,1137,1138,1144,1150,1156,1162,1168,1174,1180,1186],{},[38,1139,1140,1143],{},[41,1141,1142],{},"多 Agent 对话","：Group Chat 模式，多个 Agent 自动对话协作完成任务",[38,1145,1146,1149],{},[41,1147,1148],{},"代码执行","：内置 Docker 代码执行器，Agent 可写代码 + 运行 + 调试",[38,1151,1152,1155],{},[41,1153,1154],{},"工具调用","：自定义函数工具，Agent 自动选择和调用",[38,1157,1158,1161],{},[41,1159,1160],{},"人在回路","：Human-in-the-loop 模式，关键决策需人工确认",[38,1163,1164,1167],{},[41,1165,1166],{},"事件驱动架构","：0.4+ 重构为 async 事件驱动，支持分布式 Agent",[38,1169,1170,1173],{},[41,1171,1172],{},"可定制 Agent","：system message \u002F 工具集 \u002F 终止条件全可自定义",[38,1175,1176,1179],{},[41,1177,1178],{},"多模型支持","：OpenAI \u002F Claude \u002F Azure \u002F Ollama \u002F Gemini 等",[38,1181,1182,1185],{},[41,1183,1184],{},"Agent 可组合","：嵌套 Agent、层级 Agent、条件路由",[38,1187,1188,1191],{},[41,1189,1190],{},"可观测性","：集成 OpenTelemetry \u002F LangSmith 追踪 Agent 行为",[20,1193,95],{"id":95},[25,1195,1196],{},"完全免费、MIT 开源、商用免费。运行成本仅来自所接入的 LLM API 调用费用。",[20,1198,171],{"id":170},[163,1200,1201],{},[25,1202,176,1203],{},[41,1204,179],{},[35,1206,1207,1210,1213,1216,1223,1226],{},[38,1208,1209],{},"Group Chat 模式让多 Agent 协作真正\"自动化\"，代码 reviewer + coder + tester 角色分工清晰",[38,1211,1212],{},"Docker 代码执行器安全隔离，Agent 写的代码在沙箱中运行",[38,1214,1215],{},"0.4+ 的事件驱动架构性能提升明显，异步并发能力强",[38,1217,1218,1219,1222],{},"自定义工具集成灵活，Python 函数加 ",[195,1220,1221],{},"@user_function"," 装饰器即可",[38,1224,1225],{},"微软背书，学术认可度高，论文引用多",[38,1227,1228],{},"人在回路模式适合需要人工把关的高风险场景",[25,1230,1231],{},[41,1232,209],{},[35,1234,1235,1238,1241,1244,1247,1250,1253],{},[38,1236,1237],{},"学习曲线非常陡峭，文档虽全但概念密集，新手容易劝退",[38,1239,1240],{},"0.2 → 0.4 API 大改，迁移成本高，网上旧教程大量失效",[38,1242,1243],{},"Agent 对话容易\"跑飞\"——无限循环 \u002F 偏离主题，需仔细设计终止条件",[38,1245,1246],{},"无 GUI，调试全靠日志和 print，排查多 Agent 对话链路费时",[38,1248,1249],{},"Token 消耗大——多 Agent 对话轮次多，API 费用叠加明显",[38,1251,1252],{},"错误处理不够健壮，LLM 返回格式异常时容易崩溃",[38,1254,1255],{},"社区活跃度不如 LangChain \u002F CrewAI，遇到问题搜索不到答案",[20,1257,235],{"id":235},[237,1259,1260,1266,1273,1279,1285,1291],{},[38,1261,1262,1265],{},[195,1263,1264],{},"pip install autogen-agentchat autogen-ext","（0.4+ 新包名）",[38,1267,1268,1269,1272],{},"配置 LLM：设置 ",[195,1270,1271],{},"OPENAI_API_KEY"," 环境变量或代码内传入",[38,1274,1275,1276],{},"定义 Agent：",[195,1277,1278],{},"AssistantAgent(name=\"coder\", system_message=\"...\", model_client=...)",[38,1280,1281,1282],{},"创建 Group Chat：",[195,1283,1284],{},"RoundRobinGroupChat(agents=[agent1, agent2])",[38,1286,1287,1288],{},"发起任务：",[195,1289,1290],{},"result = await team.run(task=\"写一个贪吃蛇游戏\")",[38,1292,1293],{},"进阶：加 Docker 代码执行器 + 自定义工具 + 人在回路",[20,1295,267],{"id":267},[97,1297,1298,1313],{},[100,1299,1300],{},[103,1301,1302,1304,1306,1308,1311],{},[106,1303,276],{},[106,1305,533],{},[106,1307,538],{},[106,1309,1310],{},"LangGraph",[106,1312,284],{},[115,1314,1315,1330,1343,1360,1375,1389,1405],{},[103,1316,1317,1320,1323,1325,1327],{},[120,1318,1319],{},"形态",[120,1321,1322],{},"Python 框架",[120,1324,1322],{},[120,1326,1322],{},[120,1328,1329],{},"可视化平台",[103,1331,1332,1335,1337,1339,1341],{},[120,1333,1334],{},"上手难度",[120,1336,882],{},[120,1338,336],{},[120,1340,882],{},[120,1342,329],{},[103,1344,1345,1348,1351,1354,1357],{},[120,1346,1347],{},"多 Agent",[120,1349,1350],{},"✅ Group Chat",[120,1352,1353],{},"✅ Crew 角色",[120,1355,1356],{},"✅ 图编排",[120,1358,1359],{},"✅ 工作流",[103,1361,1362,1364,1367,1370,1372],{},[120,1363,1148],{},[120,1365,1366],{},"✅ Docker 沙箱",[120,1368,1369],{},"需自定义",[120,1371,1369],{},[120,1373,1374],{},"沙箱",[103,1376,1377,1380,1382,1385,1387],{},[120,1378,1379],{},"GUI",[120,1381,872],{},[120,1383,1384],{},"❌（有 CrewAI Studio）",[120,1386,872],{},[120,1388,347],{},[103,1390,1391,1394,1397,1400,1403],{},[120,1392,1393],{},"API 稳定性",[120,1395,1396],{},"一般（大改过）",[120,1398,1399],{},"较好",[120,1401,1402],{},"好",[120,1404,1402],{},[103,1406,1407,1410,1413,1416,1419],{},[120,1408,1409],{},"适合场景",[120,1411,1412],{},"研究 \u002F 复杂协作",[120,1414,1415],{},"业务自动化",[120,1417,1418],{},"精确流程控制",[120,1420,1421],{},"应用构建",[20,1423,411],{"id":411},[35,1425,1426,1436,1442,1448,1454,1460,1466],{},[38,1427,1428,1431,1432,1435],{},[41,1429,1430],{},"锁定版本","：0.2 和 0.4 API 不兼容，",[195,1433,1434],{},"pip install"," 时务必指定版本",[38,1437,1438,1441],{},[41,1439,1440],{},"设计终止条件","：Group Chat 不设终止条件会无限对话，设 max_turns + 终止关键词",[38,1443,1444,1447],{},[41,1445,1446],{},"Token 成本控制","：多 Agent 对话 token 消耗是单 Agent 的 3-5 倍，用 GPT-4 级模型注意费用",[38,1449,1450,1453],{},[41,1451,1452],{},"代码执行器一定要用 Docker","：直接本地执行 Agent 生成的代码有安全风险",[38,1455,1456,1459],{},[41,1457,1458],{},"别指望第一次跑通","：system message 调试 + 工具定义 + 终止条件需要反复迭代",[38,1461,1462,1465],{},[41,1463,1464],{},"错误处理要完善","：LLM 返回异常格式时手动 catch + 重试",[38,1467,1468,1471],{},[41,1469,1470],{},"不要用旧教程","：0.4+ 完全重构，网上大部分 AutoGen 教程是 0.2 版本的",[20,1473,459],{"id":458},[35,1475,1476,1479,1482,1485,1488,1491,1494,1497,1500,1503],{},[38,1477,1478],{},"✅ AI 研究者实验多 Agent 协作模式",[38,1480,1481],{},"✅ 需要代码级精细控制 Agent 行为",[38,1483,1484],{},"✅ 需要 Agent 代码执行 + 自动调试",[38,1486,1487],{},"✅ 人在回路的高风险决策场景",[38,1489,1490],{},"✅ 学术项目 \u002F 论文复现",[38,1492,1493],{},"❌ 快速原型验证（用 CrewAI，API 更简洁）",[38,1495,1496],{},"❌ 非技术用户（用 Dify \u002F Flowise）",[38,1498,1499],{},"❌ 追求 API 稳定性的生产项目（版本变动大）",[38,1501,1502],{},"❌ 需要可视化调试（无 GUI，全靠日志）",[38,1504,1505],{},"❌ 预算敏感场景（多 Agent 对话 token 消耗大）",[20,1507,495],{"id":494},[25,1509,1510,1513],{},[41,1511,1512],{},"Q: AutoGen 和 CrewAI 怎么选？","\nA: AutoGen 更底层、更灵活，适合研究和复杂多 Agent 协作实验，但学习成本高。CrewAI API 更简洁直观，角色 + 任务 + 流程的概念更易理解，适合业务自动化场景。研究选 AutoGen，做产品选 CrewAI。",[25,1515,1516,1519,1520,1523,1524,1527,1528,1531],{},[41,1517,1518],{},"Q: AutoGen 0.2 和 0.4 有什么区别？","\nA: 0.4 是完全重构版本——从同步改为异步事件驱动架构，包名从 ",[195,1521,1522],{},"pyautogen"," 改为 ",[195,1525,1526],{},"autogen-agentchat"," + ",[195,1529,1530],{},"autogen-ext","，API 全面更新。性能和扩展性大幅提升但旧代码无法直接迁移。新项目直接用 0.4+。",[25,1533,1534,1537],{},[41,1535,1536],{},"Q: 多 Agent 对话成本高吗？","\nA: 高。多 Agent 每轮对话都消耗 token，一个任务 5-10 轮对话是常态，使用 GPT-4 级模型单个任务可能花费 $0.5-2。建议开发调试用便宜模型（GPT-4o-mini），生产再切高级模型。",[25,1539,1540,1542,1543,1545],{},[41,1541,512],{},"\nA: 可以。通过 ",[195,1544,1530],{}," 的 OpenAI 兼容客户端接入 Ollama \u002F vLLM \u002F LM Studio 的本地模型端点。但本地模型能力有限，复杂多 Agent 协作效果可能不如 GPT-4 \u002F Claude。",[20,1547,526],{"id":526},[25,1549,1550,534,1552,534,1554],{},[530,1551,538],{"href":537},[530,1553,10],{"href":1054},[530,1555,1557],{"href":1556},"\u002Fcoding\u002Fapi\u002Flangfuse.html","Langfuse",[20,1559,545],{"id":545},[163,1561,1562],{},[25,1563,550],{},[35,1565,1566,1572],{},[38,1567,1568],{},[530,1569,560],{"href":1570,"rel":1571},"https:\u002F\u002Fmicrosoft.github.io\u002Fautogen",[559],[38,1573,1574],{},[530,1575,567],{"href":1576,"rel":1577},"https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fautogen",[559],{"title":569,"searchDepth":570,"depth":570,"links":1579},[1580,1581,1582,1583,1584,1585,1586,1587,1588,1589,1590],{"id":22,"depth":573,"text":23},{"id":33,"depth":573,"text":33},{"id":95,"depth":573,"text":95},{"id":170,"depth":573,"text":171},{"id":235,"depth":573,"text":235},{"id":267,"depth":573,"text":267},{"id":411,"depth":573,"text":411},{"id":458,"depth":573,"text":459},{"id":494,"depth":573,"text":495},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},"\u002Fimg\u002Ftools\u002Fautogen.webp","AutoGen 真实评测：微软开源的多 Agent 对话框架（MIT 协议），通过代码定义 Agent 角色和协作流程，支持多 Agent 对话、工具调用、代码执行。适合需要精细控制多 Agent 协作逻辑的开发者和研究团队。",[590],{},"\u002Ftools\u002Fagent\u002Fplatform\u002Fautogen",[597,598],"Free \u002F 开源（MIT）",{"power":602,"ux":573,"price":603,"cn_support":573,"stability":570},{"title":533,"description":1592},"AutoGen - 微软多 Agent 框架评测与使用 | AIHO","agent\u002Fplatform\u002Fautogen",[1603,1604],{"title":560,"url":1570},{"title":567,"url":1576},"tools\u002Fagent\u002Fplatform\u002Fautogen","微软开源多 Agent 对话框架，代码驱动 Agent 协作",[613,1608,1609,1610,1611,614],"multi-agent","framework","microsoft","python","需要代码级精细控制多 Agent 协作逻辑的研究者和高级开发者首选，微软背书 + 代码执行 + Group Chat 模式强大，但学习曲线陡峭、API 稳定性一般、无 GUI，不适合快速原型或非技术用户。","Asmg5F2Nn2KaD4XJWzqRwFA4s1iK_qUmi34M41yhcNE",{"id":1615,"title":1616,"alternatives":1617,"api_compatible":1619,"body":1620,"category":584,"chinese_friendly":603,"cover":2446,"description":2447,"domestic":587,"extension":588,"faq":15,"free":587,"github":15,"languages":2448,"lastVerified":15,"meta":2450,"models":2451,"navigation":593,"notSuitable":2457,"opensource":587,"path":2461,"pillar":595,"platforms":2462,"priceTable":2464,"pricing":2485,"published":2486,"relatedPlaybooks":15,"relatedReviews":2487,"score":2492,"self_host":587,"seo":2493,"seoTitle":2494,"slug":2495,"sources":2496,"stem":2507,"suitable":2508,"tagline":2514,"tags":2515,"updated":2520,"verdict":2521,"website":1642,"__hash__":2522},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fcoze.md","Coze",[13,625,1618],"agent\u002Fplatform\u002Fyuanqi",[],{"type":17,"value":1621,"toc":2430},[1622,1624,1665,1670,1673,1678,1681,1695,1704,1754,1763,1767,1770,1801,1809,1813,1816,1836,1840,1920,1930,1933,1936,1956,1965,1969,2005,2008,2011,2151,2167,2198,2201,2274,2276,2279,2296,2299,2328,2330,2393,2395,2422],[20,1623,23],{"id":22},[1625,1626,1631,1652],"div",{"className":1627},[1628,1629,1630],"card","p-5","my-4",[25,1632,1633,1636,1637,1651],{},[41,1634,1635],{},"一句话："," 字节跳动出品的低代码 Agent 平台，",[41,1638,1639,1640,1645,1646],{},"国内版 ",[530,1641,1644],{"href":1642,"rel":1643},"https:\u002F\u002Fwww.coze.cn",[559],"coze.cn","（中文叫\"扣子\"）+ 国际版 ",[530,1647,1650],{"href":1648,"rel":1649},"https:\u002F\u002Fwww.coze.com",[559],"coze.com"," 双轨运营。国内版深度集成飞书 \u002F 抖音生态、原生接入豆包；国际版集成 OpenAI \u002F Claude \u002F Gemini。",[25,1653,1654,1655,1658,1659,1664],{},"最大价值是 ",[41,1656,1657],{},"零代码、最快上手","——文档详细 + 中文社区活跃 + 模板多，根据 ",[530,1660,1663],{"href":1661,"rel":1662},"https:\u002F\u002Fwww.cnblogs.com\u002Fuulucias\u002Fp\u002F19449008",[559],"博客园 2026-01 选型指南"," 引用的真实案例，\"某电商公司用 Coze 搭建客服机器人，3 天上线，月成本 \u003C 1000 元\"。",[163,1666,1667],{},[25,1668,1669],{},"来源说明：本文基于 coze.cn \u002F coze.com 官方页面、docs.coze.com 文档、第三方选型评测（cnblogs \u002F besthub \u002F aibotgo）综合整理。字节产品迭代很快，价格 \u002F 功能请以最新官方页面为准。",[20,1671,1672],{"id":1672},"核心特性",[1674,1675,1677],"h3",{"id":1676},"可视化工作流最大卖点","可视化工作流（最大卖点）",[25,1679,1680],{},"Coze 把 Agent 拆成两层：",[35,1682,1683,1689],{},[38,1684,1685,1688],{},[41,1686,1687],{},"Bot \u002F 智能体","：对话式 AI，配置 prompt + 知识库 + 插件",[38,1690,1691,1694],{},[41,1692,1693],{},"工作流（Workflow）","：DAG 节点编排，可被 Bot 调用，也可独立部署",[25,1696,1697,1698,1703],{},"工作流节点类型（基于 ",[530,1699,1702],{"href":1700,"rel":1701},"https:\u002F\u002Fdeveloper.volcengine.com\u002Farticles\u002F7530117616687480851",[559],"火山引擎社区 2025 实战","）：",[35,1705,1706,1712,1718,1724,1730,1736,1742,1748],{},[38,1707,1708,1711],{},[41,1709,1710],{},"开始 \u002F 结束节点","：输入输出",[38,1713,1714,1717],{},[41,1715,1716],{},"大模型节点","：调豆包 \u002F GPT \u002F Claude 任一模型",[38,1719,1720,1723],{},[41,1721,1722],{},"代码节点","：内嵌 Python \u002F JavaScript（飞书插件常需要数据格式转换）",[38,1725,1726,1729],{},[41,1727,1728],{},"循环节点","：批量处理多条数据",[38,1731,1732,1735],{},[41,1733,1734],{},"条件节点","：分支判断",[38,1737,1738,1741],{},[41,1739,1740],{},"插件节点","：调 Coze 插件市场的工具",[38,1743,1744,1747],{},[41,1745,1746],{},"知识库节点","：RAG 检索",[38,1749,1750,1753],{},[41,1751,1752],{},"HTTP 节点","：调外部 API",[25,1755,1756,1757,1762],{},"典型用例（参考 ",[530,1758,1761],{"href":1759,"rel":1760},"https:\u002F\u002Fwww.toutiao.com\u002Farticle\u002F7469986334686315017",[559],"今日头条 涛哥讲AI 2025-02 教程","）：读飞书多维表格 → 批量调大模型转写为小红书风格 → 写回飞书。整个流程零代码完成。",[1674,1764,1766],{"id":1765},"插件市场200-官方-海量第三方","插件市场（200+ 官方 + 海量第三方）",[25,1768,1769],{},"Coze 的\"插件\"是封装好的 API 工具，比如：",[35,1771,1772,1778,1784,1790,1795],{},[38,1773,1774,1777],{},[41,1775,1776],{},"飞书多维表格","：增删改查记录（国内 toB 场景的杀手锏）",[38,1779,1780,1783],{},[41,1781,1782],{},"图像生成","：调豆包 \u002F SD \u002F DALL-E",[38,1785,1786,1789],{},[41,1787,1788],{},"联网搜索","：实时网页检索",[38,1791,1792,1794],{},[41,1793,1148],{},"：在线运行 Python",[38,1796,1797,1800],{},[41,1798,1799],{},"第三方 SaaS","：微博、抖音、bilibili、Notion……",[25,1802,1803,1808],{},[530,1804,1807],{"href":1805,"rel":1806},"https:\u002F\u002Fkouziai.github.io\u002F",[559],"扣子空间介绍"," 提到：\"插件数量突破 500 个\"——可信度待官方确认，但量级在百级别是确定的。",[1674,1810,1812],{"id":1811},"bot-商店-多平台一键发布","Bot 商店 + 多平台一键发布",[25,1814,1815],{},"发布渠道：",[35,1817,1818,1821,1824,1827,1830,1833],{},[38,1819,1820],{},"飞书机器人（一键绑）",[38,1822,1823],{},"抖音 \u002F 头条号",[38,1825,1826],{},"微信小程序 \u002F 公众号（部分需企业认证）",[38,1828,1829],{},"自定义网页嵌入",[38,1831,1832],{},"API 接口（提供 OpenAPI 风格 REST 调用）",[38,1834,1835],{},"Discord（国际版）",[1674,1837,1839],{"id":1838},"国内版-vs-国际版","国内版 vs 国际版",[97,1841,1842,1854],{},[100,1843,1844],{},[103,1845,1846,1848,1851],{},[106,1847,276],{},[106,1849,1850],{},"扣子（coze.cn）",[106,1852,1853],{},"Coze（coze.com）",[115,1855,1856,1867,1877,1887,1898,1909],{},[103,1857,1858,1861,1864],{},[120,1859,1860],{},"主力模型",[120,1862,1863],{},"豆包 Pro \u002F DeepSeek \u002F Qwen \u002F Kimi",[120,1865,1866],{},"OpenAI \u002F Claude \u002F Gemini \u002F Cohere",[103,1868,1869,1872,1875],{},[120,1870,1871],{},"飞书 \u002F 抖音 \u002F 微信集成",[120,1873,1874],{},"✅ 原生",[120,1876,872],{},[103,1878,1879,1882,1885],{},[120,1880,1881],{},"Discord \u002F Slack 集成",[120,1883,1884],{},"⚠️ 有限",[120,1886,1874],{},[103,1888,1889,1892,1895],{},[120,1890,1891],{},"数据存储位置",[120,1893,1894],{},"国内",[120,1896,1897],{},"海外",[103,1899,1900,1903,1906],{},[120,1901,1902],{},"支付",[120,1904,1905],{},"微信 \u002F 支付宝",[120,1907,1908],{},"海外信用卡",[103,1910,1911,1914,1917],{},[120,1912,1913],{},"内容合规",[120,1915,1916],{},"严格审核",[120,1918,1919],{},"宽松",[25,1921,1922,1925,1926,1929],{},[41,1923,1924],{},"实践建议","：国内 toC \u002F toB 用扣子，海外项目 \u002F 接 GPT 用 coze.com。两边账号 \u002F 工作流 ",[41,1927,1928],{},"不互通","。",[20,1931,1932],{"id":1932},"价格与运行成本",[25,1934,1935],{},"国内版（扣子）：",[35,1937,1938,1944,1950],{},[38,1939,1940,1943],{},[41,1941,1942],{},"免费版","：免费模型有日额度（豆包 lite 等），适合个人玩 \u002F Demo",[38,1945,1946,1949],{},[41,1947,1948],{},"专业版","：按调用计费，模型 + 高并发，单 token 价比直连 API 略贵但省事",[38,1951,1952,1955],{},[41,1953,1954],{},"企业版","：议价，含 VPC、私有化（限定场景）、SLA",[25,1957,1958,1959,1964],{},"国际版（coze.com）的定价模式据 ",[530,1960,1963],{"href":1961,"rel":1962},"https:\u002F\u002Fdocs.coze.com\u002F",[559],"官方文档"," 描述：\"按你访问和使用的功能分别计费，每个功能有自己的计费模型\"——目前没有简单的\"$X\u002F月\"档位，类似按 token \u002F 工具调用的 metered billing。",[20,1966,1968],{"id":1967},"上手-10-分钟","上手 10 分钟",[237,1970,1971,1984,1987,1990,1993,1996,1999,2002],{},[38,1972,1973,1974,1978,1979,1983],{},"打开 ",[530,1975,1977],{"href":1642,"rel":1976},[559],"www.coze.cn","（国内）或 ",[530,1980,1982],{"href":1648,"rel":1981},[559],"www.coze.com","（国际），用飞书 \u002F 抖音账号 \u002F Google 账号登录",[38,1985,1986],{},"左侧\"工作空间\" → \"+创建 Bot\"，起个名字",[38,1988,1989],{},"选模型（国内推荐豆包 Pro，国际推 Claude Sonnet 4）",[38,1991,1992],{},"写 Bot 角色 prompt",[38,1994,1995],{},"可选：上传 PDF \u002F 文档建知识库",[38,1997,1998],{},"测试一下对话效果",[38,2000,2001],{},"右上\"发布\" → 选渠道（飞书 \u002F 抖音 \u002F API \u002F Web）",[38,2003,2004],{},"拿到调用 URL \u002F 飞书机器人 webhook",[25,2006,2007],{},"进阶：在\"资源库\"创建工作流，拖节点 → 调试 → 在 Bot 里\"添加工作流\"引用。",[20,2009,2010],{"id":2010},"与同类怎么选",[97,2012,2013,2037],{},[100,2014,2015],{},[103,2016,2017,2019,2021,2026,2031],{},[106,2018,276],{},[106,2020,1616],{},[106,2022,2023],{},[530,2024,284],{"href":2025},"\u002Fagent\u002Fplatform\u002Fdify.html",[106,2027,2028],{},[530,2029,839],{"href":2030},"\u002Fagent\u002Fplatform\u002Ffastgpt.html",[106,2032,2033],{},[530,2034,2036],{"href":2035},"\u002Fagent\u002Fplatform\u002Fyuanqi.html","元器 yuanqi",[115,2038,2039,2052,2067,2082,2098,2110,2124,2138],{},[103,2040,2041,2044,2046,2048,2050],{},[120,2042,2043],{},"开源",[120,2045,872],{},[120,2047,347],{},[120,2049,347],{},[120,2051,872],{},[103,2053,2054,2057,2060,2062,2064],{},[120,2055,2056],{},"私有部署",[120,2058,2059],{},"⚠️ 仅企业版",[120,2061,347],{},[120,2063,347],{},[120,2065,2066],{},"⚠️",[103,2068,2069,2071,2074,2077,2079],{},[120,2070,1334],{},[120,2072,2073],{},"★ 最简单",[120,2075,2076],{},"★★★",[120,2078,2076],{},[120,2080,2081],{},"★★",[103,2083,2084,2087,2090,2093,2096],{},[120,2085,2086],{},"工作流编排",[120,2088,2089],{},"★★★★☆",[120,2091,2092],{},"★★★★★",[120,2094,2095],{},"★★★☆☆",[120,2097,2095],{},[103,2099,2100,2102,2104,2106,2108],{},[120,2101,877],{},[120,2103,2095],{},[120,2105,2089],{},[120,2107,2092],{},[120,2109,2095],{},[103,2111,2112,2115,2117,2119,2121],{},[120,2113,2114],{},"字节生态",[120,2116,2092],{},[120,2118,872],{},[120,2120,872],{},[120,2122,2123],{},"❌（腾讯系）",[103,2125,2126,2129,2132,2134,2136],{},[120,2127,2128],{},"插件市场",[120,2130,2131],{},"★★★★★ 200+",[120,2133,2095],{},[120,2135,2095],{},[120,2137,2095],{},[103,2139,2140,2143,2145,2147,2149],{},[120,2141,2142],{},"中文社区",[120,2144,2092],{},[120,2146,2089],{},[120,2148,2089],{},[120,2150,2089],{},[25,2152,2153,2156,2157,2162,2163,1703],{},[41,2154,2155],{},"怎么选","（综合 ",[530,2158,2161],{"href":2159,"rel":2160},"https:\u002F\u002Fwww.besthub.dev\u002Farticles\u002Fcoze-vs-dify-vs-fastgpt-which-ai-agent-platform-fits-your-needs-fa59cf97b798",[559],"BestHub 2025-07"," 和 ",[530,2164,2166],{"href":1661,"rel":2165},[559],"博客园 2026-01",[35,2168,2169,2175,2183,2190],{},[38,2170,2171,2174],{},[41,2172,2173],{},"快速验证 \u002F 不懂代码 \u002F 1-2 天出原型"," → Coze",[38,2176,2177,2180,2181],{},[41,2178,2179],{},"数据安全要求高 \u002F 复杂业务流程 \u002F 有技术团队"," → ",[530,2182,284],{"href":2025},[38,2184,2185,2180,2188],{},[41,2186,2187],{},"核心场景就是企业知识库 QA",[530,2189,839],{"href":2030},[38,2191,2192,2180,2195],{},[41,2193,2194],{},"QQ \u002F 微信生态 + 腾讯系",[530,2196,2197],{"href":2035},"元器",[20,2199,2200],{"id":2200},"避坑清单",[35,2202,2203,2213,2219,2234,2244,2250,2262,2268],{},[38,2204,2205,2208,2209,2212],{},[41,2206,2207],{},"国内版 vs 国际版的\"双账号陷阱\"","：扣子（coze.cn）和 Coze（coze.com）是",[41,2210,2211],{},"两套独立系统","，账号、Bot、工作流不互通；想\"国内调通后搬到海外\"需要重新搭",[38,2214,2215,2218],{},[41,2216,2217],{},"专业版按调用计费容易超预算","：上线前一定在测试环境跑量估算月成本，否则爆款 Bot 一夜烧爆账户",[38,2220,2221,2224,2225,2228,2229,2233],{},[41,2222,2223],{},"飞书多维表格插件数据格式坑","：写入多维表格需要 ",[195,2226,2227],{},"Array\u003CObject>"," 格式，代码节点要做转换（参考 ",[530,2230,2232],{"href":1700,"rel":2231},[559],"火山引擎 2025 教程","）",[38,2235,2236,2239,2240,2243],{},[41,2237,2238],{},"工作流读取飞书表格默认 20 条","：要改 ",[195,2241,2242],{},"page_size","，最大 500 条；超过 500 要分页或循环",[38,2245,2246,2249],{},[41,2247,2248],{},"运行超时","：单工作流执行有时间上限，记录条数 > 50 时建议在 Bot 里\"异步\"调用，不要直接走工作流",[38,2251,2252,2255,2256,2258,2259,2261],{},[41,2253,2254],{},"企业版\"私有化\"是有限的","：完全数据不出网仍建议 ",[530,2257,284],{"href":2025}," \u002F ",[530,2260,839],{"href":2030}," 自托管",[38,2263,2264,2267],{},[41,2265,2266],{},"国际版接 Claude \u002F GPT 需要 BYOK","：自己绑海外信用卡，平台不代付",[38,2269,2270,2273],{},[41,2271,2272],{},"审核合规","：国内版对 prompt \u002F 输出有内容审核，金融 \u002F 医疗 \u002F 政治话题可能被拦",[20,2275,459],{"id":458},[25,2277,2278],{},"✅ 适合：",[35,2280,2281,2284,2287,2290,2293],{},[38,2282,2283],{},"产品 \u002F 运营 \u002F 非技术人员快速做 Bot",[38,2285,2286],{},"在飞书 \u002F 抖音 \u002F 头条生态内做集成",[38,2288,2289],{},"个人副业（小红书账号批量内容生成等）",[38,2291,2292],{},"中小企业客服 Bot（3 天上线）",[38,2294,2295],{},"想用豆包 \u002F DeepSeek 国产模型的人",[25,2297,2298],{},"❌ 不适合：",[35,2300,2301,2304,2307,2312,2319],{},[38,2302,2303],{},"金融 \u002F 政府 \u002F 医疗（数据敏感，需自托管）",[38,2305,2306],{},"复杂业务系统深度集成（自由度不够）",[38,2308,2309,2310,2233],{},"反感字节生态（去 ",[530,2311,284],{"href":2025},[38,2313,2314,2315,2258,2317,2233],{},"希望开源 \u002F 完全自主可控（去 ",[530,2316,284],{"href":2025},[530,2318,839],{"href":2030},[38,2320,2321,2322,2258,2324,2327],{},"海外 toB SaaS 产品后端（",[530,2323,284],{"href":2025},[530,2325,281],{"href":2326},"\u002Fagent\u002Fplatform\u002Flangflow.html"," 更合适）",[20,2329,526],{"id":526},[35,2331,2332,2346,2363,2382],{},[38,2333,2334,2335,2258,2337,2258,2339,2258,2341,2258,2343],{},"同类对比：",[530,2336,284],{"href":2025},[530,2338,839],{"href":2030},[530,2340,2197],{"href":2035},[530,2342,281],{"href":2326},[530,2344,287],{"href":2345},"\u002Fagent\u002Fplatform\u002Fn8n.html",[38,2347,2348,2349,2258,2353,2258,2356,2258,2360],{},"概念：",[530,2350,2352],{"href":2351},"\u002Fwiki\u002Fai-agent.html","AI Agent",[530,2354,341],{"href":2355},"\u002Fwiki\u002Frag.html",[530,2357,2359],{"href":2358},"\u002Fwiki\u002Ffunction-calling.html","Function Calling",[530,2361,2362],{"href":2351},"Multi-Agent",[38,2364,2365,2366,2258,2370,2258,2374,2258,2378],{},"模型：",[530,2367,2369],{"href":2368},"\u002Fmodels\u002Fdoubao-1-5-pro.html","豆包 Doubao",[530,2371,2373],{"href":2372},"\u002Fmodels\u002Fdeepseek-v3.html","DeepSeek-V3",[530,2375,2377],{"href":2376},"\u002Fmodels\u002Fqwen-3.html","Qwen3",[530,2379,2381],{"href":2380},"\u002Fmodels\u002Fglm-5.2.html","GLM-5.2",[38,2383,2384,2385,2258,2389],{},"进阶：",[530,2386,2388],{"href":2387},"\u002Fwiki\u002Fprompt-engineering.html","Prompt Engineering",[530,2390,2392],{"href":2391},"\u002Fwiki\u002Fcontext-engineering.html","Context Engineering",[20,2394,545],{"id":545},[35,2396,2397,2403,2409,2416,2419],{},[38,2398,2399,2400],{},"国内版：",[530,2401,1642],{"href":1642,"rel":2402},[559],[38,2404,2405,2406],{},"国际版：",[530,2407,1648],{"href":1648,"rel":2408},[559],[38,2410,2411,2412],{},"官方文档：",[530,2413,2414],{"href":2414,"rel":2415},"https:\u002F\u002Fdocs.coze.com",[559],[38,2417,2418],{},"第三方选型评测：cnblogs.com \u002F besthub.dev \u002F aibotgo.net",[38,2420,2421],{},"实战教程：火山引擎社区、今日头条 涛哥讲AI",[25,2423,2424,2425,2429],{},"本卡片由 AIHO 编辑部根据官方公开资料与第三方评测整理。所有事实点均标注来源；如发现价格 \u002F 功能 \u002F 渠道与最新官方信息不一致，请通过 ",[530,2426,2428],{"href":2427},"mailto:hello@aiho.net","反馈邮箱"," 反馈。",{"title":569,"searchDepth":570,"depth":570,"links":2431},[2432,2433,2439,2440,2441,2442,2443,2444,2445],{"id":22,"depth":573,"text":23},{"id":1672,"depth":573,"text":1672,"children":2434},[2435,2436,2437,2438],{"id":1676,"depth":570,"text":1677},{"id":1765,"depth":570,"text":1766},{"id":1811,"depth":570,"text":1812},{"id":1838,"depth":570,"text":1839},{"id":1932,"depth":573,"text":1932},{"id":1967,"depth":573,"text":1968},{"id":2010,"depth":573,"text":2010},{"id":2200,"depth":573,"text":2200},{"id":458,"depth":573,"text":459},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},"\u002Fimg\u002Ftools\u002Fcoze.webp","Coze 扣子 2026 真实评测：字节低代码 AI Agent 平台，支持可视化工作流、插件市场、知识库和 Bot 发布。本文对比 Dify、FastGPT、n8n，梳理国内版\u002F国际版差异、价格、适合场景和避坑建议。",[2449,590],"zh",{},[2452,2453,2454,2455,2456],"doubao-pro","deepseek-v3","qwen-max","gpt-4o (国际版)","claude (国际版)",[2458,2459,2460],"需要私有部署、数据不出网（去 Dify \u002F FastGPT）","需要深度定制（去 LangGraph \u002F n8n）","对字节生态依赖反感","\u002Ftools\u002Fagent\u002Fplatform\u002Fcoze",[2463],"web",[2465,2471,2476,2480],{"plan":2466,"price":2467,"limit":2468,"cn_pay":2469,"note":2470},"免费版（扣子）","¥0","免费模型限额 + 基础功能","—","个人试水 \u002F 1-2 天 MVP",{"plan":1948,"price":2472,"limit":2473,"cn_pay":2474,"note":2475},"按调用计费","高级模型 + 高并发 + 商用授权","✅ 微信\u002F支付宝","上量后切",{"plan":1954,"price":2477,"limit":2478,"cn_pay":347,"note":2479},"议价","VPC、合规、SLA、私有化（部分）","B 端落地",{"plan":2481,"price":2482,"cn_pay":2483,"note":2484,"limit":2469},"国际版 (coze.com)","免费起步 + 用量计费","需海外卡","可接 GPT\u002FClaude\u002FGemini","免费档 \u002F 专业版按调用计费 \u002F 企业版议价","2026-06-18",[2488,2489,2490,2491],"coze-deep-review","coze-vs-dify","dify-deep-review","fastgpt-deep-review",{"power":602,"ux":603,"price":602,"cn_support":603,"stability":602},{"title":1616,"description":2447},"Coze 扣子评测 2026：字节 AI Agent 平台，对比 Dify","agent\u002Fplatform\u002Fcoze",[2497,2499,2501,2503,2505],{"title":2498,"url":1642},"Coze 国内版（扣子）",{"title":2500,"url":1648},"Coze 国际版",{"title":2502,"url":2414},"Coze 官方文档",{"title":2504,"url":1661},"Coze vs Dify vs FastGPT 选型 2026",{"title":2506,"url":2159},"BestHub 三平台对比","tools\u002Fagent\u002Fplatform\u002Fcoze",[2509,2510,2511,2512,2513],"想 1 小时做出一个 Bot 的产品 \u002F 运营","需要发布到飞书 \u002F 微信 \u002F 抖音的 Bot","工作流可视化编排（不想写代码）","需要批量调用国内大模型 + 飞书多维表格的工作流","C 端 \u002F 轻量 toB 场景","字节出品的 Agent 搭建平台，国内 \u002F 国际双版本",[613,2516,2517,2518,2519,616],"low-code","workflow","bot-marketplace","bytedance","2026-06-24","想最快做出一个能用的 Bot，从 Coze 起步。要私有部署或开源协作，去 Dify \u002F FastGPT。","necnnd5prSTfssbiPIQOZWkZ3WDKHK3nbAKz7ZhN7uc",{"id":2524,"title":538,"alternatives":2525,"api_compatible":15,"body":2526,"category":584,"chinese_friendly":573,"cover":3004,"description":3005,"domestic":587,"extension":588,"faq":15,"free":587,"github":2989,"languages":3006,"lastVerified":591,"meta":3007,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":3008,"pillar":595,"platforms":3009,"priceTable":15,"pricing":3010,"published":600,"relatedPlaybooks":15,"relatedReviews":15,"score":3011,"self_host":587,"seo":3012,"seoTitle":3013,"slug":1119,"sources":3014,"stem":3017,"suitable":15,"tagline":3018,"tags":3019,"updated":591,"verdict":3020,"website":2983,"__hash__":3021},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fcrewai.md",[12,1601,14],{"type":17,"value":2527,"toc":2991},[2528,2530,2533,2536,2538,2593,2595,2638,2642,2644,2650,2673,2677,2700,2702,2739,2741,2853,2855,2903,2905,2936,2938,2944,2950,2955,2961,2963,2971,2973,2977],[20,2529,23],{"id":22},[25,2531,2532],{},"CrewAI 是开源多 Agent 协作框架（MIT 协议），用 Python 代码定义 Agent 角色（Role）、目标（Goal）、工具（Tools），组合成 Crew 执行任务序列。核心概念清晰——Agent 负责做事、Task 定义做什么、Crew 编排怎么协作。支持顺序 \u002F 层级 \u002F 自定义流程，内置 50+ 工具集成。CrewAI Enterprise 提供云端托管 + 可视化监控。",[25,2534,2535],{},"适合：需要多 Agent 自动化工作流的开发者、Python 技术栈团队、快速原型验证多 Agent 方案。不适合：非技术用户（用 Dify）、需要极复杂条件分支流程（用 LangGraph）、需要 GUI 可视化编排（用 Flowise）。",[20,2537,33],{"id":33},[35,2539,2540,2546,2552,2558,2564,2570,2576,2581,2587],{},[38,2541,2542,2545],{},[41,2543,2544],{},"角色定义","：Agent = Role + Goal + Backstory + Tools，角色人设驱动行为",[38,2547,2548,2551],{},[41,2549,2550],{},"任务编排","：Task 定义具体任务 + 期望输出 + 分配 Agent",[38,2553,2554,2557],{},[41,2555,2556],{},"Crew 编排","：顺序执行 \u002F 层级管理 \u002F 自定义流程三种模式",[38,2559,2560,2563],{},[41,2561,2562],{},"工具集成","：内置 SerperDev \u002F Firecrawl \u002F FileRead \u002F WebScraper 等 50+ 工具",[38,2565,2566,2569],{},[41,2567,2568],{},"流程控制","：支持任务间依赖、条件路由、输出传递",[38,2571,2572,2575],{},[41,2573,2574],{},"记忆系统","：Short-term \u002F Long-term \u002F Entity Memory，Agent 可跨任务记忆",[38,2577,2578,2580],{},[41,2579,1178],{},"：OpenAI \u002F Claude \u002F Gemini \u002F Ollama \u002F 任意 LiteLLM 兼容模型",[38,2582,2583,2586],{},[41,2584,2585],{},"CrewAI Enterprise","：云端托管 + 可视化 Crew 监控 + 团队协作",[38,2588,2589,2592],{},[41,2590,2591],{},"输出结构化","：支持 Pydantic 模型定义输出格式",[20,2594,95],{"id":95},[97,2596,2597,2607],{},[100,2598,2599],{},[103,2600,2601,2603,2605],{},[106,2602,108],{},[106,2604,95],{},[106,2606,113],{},[115,2608,2609,2618,2628],{},[103,2610,2611,2613,2615],{},[120,2612,122],{},[120,2614,125],{},[120,2616,2617],{},"完整框架，MIT 协议，本地运行",[103,2619,2620,2622,2625],{},[120,2621,2585],{},[120,2623,2624],{},"$49\u002F月起",[120,2626,2627],{},"云端托管 + 可视化监控 + API",[103,2629,2630,2633,2635],{},[120,2631,2632],{},"Enterprise+",[120,2634,158],{},[120,2636,2637],{},"SSO \u002F 私有部署 \u002F 专属支持",[163,2639,2640],{},[25,2641,167],{},[20,2643,171],{"id":170},[163,2645,2646],{},[25,2647,176,2648],{},[41,2649,179],{},[35,2651,2652,2655,2658,2661,2664,2667,2670],{},[38,2653,2654],{},"API 设计非常直观，Agent + Task + Crew 三件套 10 分钟上手",[38,2656,2657],{},"角色人设（Backstory）确实影响 Agent 行为，写好 backstory 效果提升明显",[38,2659,2660],{},"层级模式（hierarchical）下 Manager Agent 自动分配任务，适合复杂场景",[38,2662,2663],{},"内置工具丰富，SerperDev 搜索 + Firecrawl 爬虫开箱即用",[38,2665,2666],{},"Memory 系统让 Agent 跨任务保持上下文，长流程不丢信息",[38,2668,2669],{},"Pydantic 结构化输出对后续处理非常友好",[38,2671,2672],{},"CrewAI Enterprise 的可视化监控能看到每个 Agent 的思考过程",[25,2674,2675],{},[41,2676,209],{},[35,2678,2679,2682,2685,2688,2691,2694,2697],{},[38,2680,2681],{},"Agent 偶尔\"不听话\"——偏离角色设定、重复执行、跳过任务",[38,2683,2684],{},"复杂流程的调试困难，错误信息不够清晰",[38,2686,2687],{},"Token 消耗不小——多 Agent + 多轮对话 + Memory 存储",[38,2689,2690],{},"开源版无 GUI，全靠日志调试，Enterprise 版才有可视化",[38,2692,2693],{},"版本迭代快，API 偶有 breaking changes",[38,2695,2696],{},"层级模式的 Manager Agent 判断不稳定，有时分配不合理",[38,2698,2699],{},"中文 system prompt 效果不如英文，建议角色定义用英文",[20,2701,235],{"id":235},[237,2703,2704,2710,2715,2720,2726,2732],{},[38,2705,2706,2709],{},[195,2707,2708],{},"pip install crewai crewai-tools","（建议用 uv 管理虚拟环境）",[38,2711,1268,2712,2714],{},[195,2713,1271],{}," 或在代码中指定 model",[38,2716,1275,2717],{},[195,2718,2719],{},"Agent(role=\"研究员\", goal=\"搜集信息\", tools=[search_tool])",[38,2721,2722,2723],{},"定义 Task：",[195,2724,2725],{},"Task(description=\"调研XX趋势\", agent=researcher, expected_output=\"报告\")",[38,2727,2728,2729],{},"组建 Crew：",[195,2730,2731],{},"Crew(agents=[researcher, writer], tasks=[task1, task2], process=Process.sequential)",[38,2733,2734,2735,2738],{},"启动：",[195,2736,2737],{},"result = crew.kickoff()","，查看结果 + 日志",[20,2740,267],{"id":267},[97,2742,2743,2757],{},[100,2744,2745],{},[103,2746,2747,2749,2751,2753,2755],{},[106,2748,276],{},[106,2750,538],{},[106,2752,533],{},[106,2754,1310],{},[106,2756,284],{},[115,2758,2759,2771,2788,2800,2812,2826,2839],{},[103,2760,2761,2763,2765,2767,2769],{},[120,2762,1334],{},[120,2764,329],{},[120,2766,882],{},[120,2768,882],{},[120,2770,851],{},[103,2772,2773,2776,2779,2782,2785],{},[120,2774,2775],{},"API 设计",[120,2777,2778],{},"优雅直观",[120,2780,2781],{},"底层灵活",[120,2783,2784],{},"图模型",[120,2786,2787],{},"可视化",[103,2789,2790,2792,2794,2796,2798],{},[120,2791,1347],{},[120,2793,1353],{},[120,2795,1350],{},[120,2797,1356],{},[120,2799,1359],{},[103,2801,2802,2804,2806,2808,2810],{},[120,2803,1148],{},[120,2805,1369],{},[120,2807,1366],{},[120,2809,1369],{},[120,2811,1374],{},[103,2813,2814,2816,2819,2822,2824],{},[120,2815,2574],{},[120,2817,2818],{},"✅ 内置",[120,2820,2821],{},"有限",[120,2823,353],{},[120,2825,347],{},[103,2827,2828,2830,2833,2835,2837],{},[120,2829,1379],{},[120,2831,2832],{},"Enterprise 版",[120,2834,872],{},[120,2836,872],{},[120,2838,347],{},[103,2840,2841,2843,2845,2848,2851],{},[120,2842,1409],{},[120,2844,1415],{},[120,2846,2847],{},"研究",[120,2849,2850],{},"精确流程",[120,2852,1421],{},[20,2854,411],{"id":411},[35,2856,2857,2863,2869,2875,2881,2887,2897],{},[38,2858,2859,2862],{},[41,2860,2861],{},"Backstory 认真写","：角色人设直接影响 Agent 行为质量，模糊描述 = 模糊行为",[38,2864,2865,2868],{},[41,2866,2867],{},"expected_output 必填","：不定义预期输出，Agent 容易跑偏",[38,2870,2871,2874],{},[41,2872,2873],{},"控制 Agent 数量","：3-5 个 Agent 最佳，超过 8 个协调成本急升",[38,2876,2877,2880],{},[41,2878,2879],{},"Memory 按需开启","：Long-term Memory 会累积 token 消耗，简单任务关掉",[38,2882,2883,2886],{},[41,2884,2885],{},"调试用 verbose=True","：开启详细日志看 Agent 思考过程，定位问题",[38,2888,2889,2892,2893,2896],{},[41,2890,2891],{},"版本锁定","：",[195,2894,2895],{},"pip install crewai==x.x.x","，迭代快别用 latest",[38,2898,2899,2902],{},[41,2900,2901],{},"层级模式慎用","：Manager Agent 不稳定，简单场景用 sequential 更可靠",[20,2904,459],{"id":458},[35,2906,2907,2910,2913,2916,2919,2922,2924,2927,2930,2933],{},[38,2908,2909],{},"✅ Python 开发者快速构建多 Agent 工作流",[38,2911,2912],{},"✅ 内容生产流水线（调研 → 写作 → 审核）",[38,2914,2915],{},"✅ 自动化研究 \u002F 数据收集 \u002F 报告生成",[38,2917,2918],{},"✅ 需要角色分工的协作场景",[38,2920,2921],{},"✅ 快速原型验证多 Agent 方案",[38,2923,1496],{},[38,2925,2926],{},"❌ 需要极复杂条件分支流程（用 LangGraph）",[38,2928,2929],{},"❌ 需要精细控制 Agent 对话轮次（用 AutoGen）",[38,2931,2932],{},"❌ 需要免费 GUI 可视化监控（开源版无 GUI）",[38,2934,2935],{},"❌ 预算敏感的高频调用场景（多 Agent token 消耗大）",[20,2937,495],{"id":494},[25,2939,2940,2943],{},[41,2941,2942],{},"Q: CrewAI 和 AutoGen 怎么选？","\nA: CrewAI 上手更快，Agent + Task + Crew 概念直观，适合业务自动化和快速原型。AutoGen 更底层灵活，Group Chat + 代码执行 + 事件驱动适合研究和复杂协作。做产品选 CrewAI，做研究选 AutoGen。",[25,2945,2946,2949],{},[41,2947,2948],{},"Q: 开源版和 Enterprise 版差别大吗？","\nA: 开源版框架功能完整，能跑所有 Agent \u002F Task \u002F Crew。Enterprise 版主要多了云端托管（免运维）、可视化监控（看 Agent 思考过程）、团队协作和 API 服务。如果只是本地跑 Agent 工作流，开源版够用。",[25,2951,2952,2954],{},[41,2953,512],{},"\nA: 可以。CrewAI 基于 LiteLLM，支持 Ollama \u002F vLLM \u002F LM Studio 等本地模型。但本地模型能力有限，角色扮演和工具调用效果可能不如 GPT-4 \u002F Claude。建议开发用便宜模型，生产用高级模型。",[25,2956,2957,2960],{},[41,2958,2959],{},"Q: Agent 总是跑偏怎么办？","\nA: 三步排查：1）检查 Backstory 是否足够具体；2）确认 expected_output 定义清晰；3）开启 verbose=True 看思考过程定位偏移点。复杂任务拆成更小的 Task，每个 Task 目标单一明确。",[20,2962,526],{"id":526},[25,2964,2965,534,2967,534,2969],{},[530,2966,533],{"href":532},[530,2968,10],{"href":1054},[530,2970,1557],{"href":1556},[20,2972,545],{"id":545},[163,2974,2975],{},[25,2976,550],{},[35,2978,2979,2985],{},[38,2980,2981],{},[530,2982,560],{"href":2983,"rel":2984},"https:\u002F\u002Fcrewai.com",[559],[38,2986,2987],{},[530,2988,567],{"href":2989,"rel":2990},"https:\u002F\u002Fgithub.com\u002FcrewAIInc\u002FcrewAI",[559],{"title":569,"searchDepth":570,"depth":570,"links":2992},[2993,2994,2995,2996,2997,2998,2999,3000,3001,3002,3003],{"id":22,"depth":573,"text":23},{"id":33,"depth":573,"text":33},{"id":95,"depth":573,"text":95},{"id":170,"depth":573,"text":171},{"id":235,"depth":573,"text":235},{"id":267,"depth":573,"text":267},{"id":411,"depth":573,"text":411},{"id":458,"depth":573,"text":459},{"id":494,"depth":573,"text":495},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},"\u002Fimg\u002Ftools\u002Fcrewai.webp","CrewAI 真实评测：开源多 Agent 协作框架（MIT 协议），用 Python 代码定义 Agent 角色、任务和协作流程。支持角色分工、任务编排、工具集成，适合需要构建多 Agent 自动化工作流的开发者和企业团队。",[590],{},"\u002Ftools\u002Fagent\u002Fplatform\u002Fcrewai",[597,598],"Free \u002F 开源（MIT）\u002F Enterprise",{"power":602,"ux":570,"price":603,"cn_support":573,"stability":570},{"title":538,"description":3005},"CrewAI - 多 Agent 协作框架评测与使用 | AIHO",[3015,3016],{"title":560,"url":2983},{"title":567,"url":2989},"tools\u002Fagent\u002Fplatform\u002Fcrewai","多 Agent 协作框架，Python 代码定义角色和任务",[613,1608,1609,1611,614],"需要用 Python 快速构建多 Agent 自动化工作流的开发者首选，角色 + 任务 + Crew 的概念直观易学、API 设计优雅，但 Agent 对话可控性和稳定性不如 AutoGen，复杂流程编排需配合 LangGraph。","_O4h6a46IiWmQgKW4ommaTGIrJPA8tvgaA2kCNf0aa0",{"id":3023,"title":284,"alternatives":3024,"api_compatible":15,"body":3025,"category":584,"chinese_friendly":602,"cover":4013,"description":4014,"domestic":587,"extension":588,"faq":15,"free":587,"github":3304,"languages":4015,"lastVerified":15,"meta":4017,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":4018,"pillar":595,"platforms":4019,"priceTable":4020,"pricing":4037,"published":2486,"relatedPlaybooks":15,"relatedReviews":4038,"score":4039,"self_host":593,"seo":4040,"seoTitle":4041,"slug":13,"sources":4042,"stem":4054,"suitable":15,"tagline":4055,"tags":4056,"updated":2520,"verdict":4059,"website":3958,"__hash__":4060},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fdify.md",[2495,625,14,12],{"type":17,"value":3026,"toc":3991},[3027,3029,3047,3052,3054,3058,3061,3128,3131,3135,3138,3152,3168,3172,3175,3186,3190,3198,3209,3213,3216,3218,3222,3231,3289,3296,3300,3308,3371,3374,3394,3398,3401,3460,3466,3468,3562,3565,3594,3596,3724,3736,3771,3773,3845,3847,3849,3869,3871,3900,3902,3949,3951,3982,3987],[20,3028,23],{"id":22},[1625,3030,3032,3037],{"className":3031},[1628,1629,1630],[25,3033,3034,3036],{},[41,3035,1635],{}," Dify 是开源 LLMOps 平台的事实标准。GitHub 13 万 star、累计 100 万+ 生产 app（据 chatforest.com 2026 评测引用 Dify 官方数据），把\"可视化工作流编排 + RAG 知识库 + Agent + MCP 协议\"打包成一个 Docker Compose 能跑起来的东西。",[25,3038,1654,3039,3042,3043,3046],{},[41,3040,3041],{},"完全开源 + 模型不挑食","——同一个工作流里同时调 OpenAI、Anthropic、Ollama 本地、DeepSeek、Qwen 都行。代价是部署比 ",[530,3044,1616],{"href":3045},"\u002Fagent\u002Fplatform\u002Fcoze.html"," 折腾，新手得读 1-2 小时文档。",[163,3048,3049],{},[25,3050,3051],{},"来源说明：本文基于 docs.dify.ai 官方文档、langgenius\u002Fdify GitHub 仓库、第三方评测（besthub.dev \u002F chatforest.com \u002F joshuaopolko.com \u002F zhihu 知名专栏）综合归纳。版本号会变，部署要求请以官方最新文档为准。",[20,3053,1672],{"id":1672},[1674,3055,3057],{"id":3056},"可视化工作流chatflow-workflow","可视化工作流（Chatflow + Workflow）",[25,3059,3060],{},"Dify 把 LLM 应用拆成两种\"应用类型\"：",[97,3062,3063,3075],{},[100,3064,3065],{},[103,3066,3067,3070,3072],{},[106,3068,3069],{},"类型",[106,3071,1409],{},[106,3073,3074],{},"编排范式",[115,3076,3077,3090,3102,3115],{},[103,3078,3079,3084,3087],{},[120,3080,3081],{},[41,3082,3083],{},"Chatbot",[120,3085,3086],{},"简单对话机器人",[120,3088,3089],{},"prompt + tools",[103,3091,3092,3096,3099],{},[120,3093,3094],{},[41,3095,358],{},[120,3097,3098],{},"自主多步任务",[120,3100,3101],{},"ReAct \u002F Function Calling",[103,3103,3104,3109,3112],{},[120,3105,3106],{},[41,3107,3108],{},"Chatflow",[120,3110,3111],{},"对话型工作流（多轮 + 分支）",[120,3113,3114],{},"节点 DAG，带聊天上下文",[103,3116,3117,3122,3125],{},[120,3118,3119],{},[41,3120,3121],{},"Workflow",[120,3123,3124],{},"单次输入→输出（API 模式）",[120,3126,3127],{},"节点 DAG，无对话状态",[25,3129,3130],{},"节点类型覆盖：LLM、知识检索、HTTP 请求、代码执行（Python \u002F JS）、条件分支、迭代、变量聚合、参数提取、问题分类——满足\"用拖拽实现可观测的 LLM pipeline\"。",[1674,3132,3134],{"id":3133},"rag-知识库","RAG 知识库",[25,3136,3137],{},"内置完整 RAG 链路：",[237,3139,3140,3143,3146,3149],{},[38,3141,3142],{},"上传文档（PDF \u002F Word \u002F Markdown \u002F 网页）",[38,3144,3145],{},"自动分块 + embedding（可配置分段策略和 embedding 模型）",[38,3147,3148],{},"混合检索（向量 + 全文 + 重排）",[38,3150,3151],{},"引用溯源（回答末尾自动附原文片段）",[25,3153,3154,3155,3160,3161,3164,3165,3167],{},"注意：根据 ",[530,3156,3159],{"href":3157,"rel":3158},"https:\u002F\u002Fzhuanlan.zhihu.com\u002Fp\u002F1887141987838309480",[559],"知乎 LLM 实战笔记 2025-03 对比"," 的实测，Dify ",[41,3162,3163],{},"社区版默认是基础语义检索","，企业版才解锁多路召回 + 重排。RAG 极致精度场景仍推荐 ",[530,3166,839],{"href":2030},"（实测准确率高 10+ 个百分点），Dify 胜在工作流而非纯 RAG。",[1674,3169,3171],{"id":3170},"模型生态40-提供商","模型生态：40+ 提供商",[25,3173,3174],{},"Dify 通过插件市场接入主流模型——OpenAI、Anthropic、Google Gemini、Azure、AWS Bedrock、Cohere、xAI、DeepSeek、Qwen、智谱、文心、豆包、月之暗面、Ollama、LM Studio、Replicate、Together AI、OpenRouter……几乎你能数出来的 LLM 提供商都在。",[25,3176,3177,3178,3180,3181,3185],{},"国产模型原生支持（不像 ",[530,3179,839],{"href":2030}," 需要 ",[530,3182,3184],{"href":3183},"\u002Fcoding\u002Fapi\u002Fone-api.html","OneAPI"," 中转），是 Dify 在国内 toB 场景流行的关键。",[1674,3187,3189],{"id":3188},"mcp-协议支持","MCP 协议支持",[25,3191,3192,3193,3197],{},"Dify 较早接入了 ",[530,3194,3196],{"href":3195},"\u002Fwiki\u002Fmcp.html","MCP（Model Context Protocol）","，工作流可以直接调 MCP Server 暴露的 tools。意味着你可以让 Dify 工作流：",[35,3199,3200,3203,3206],{},[38,3201,3202],{},"通过 MCP 调本地 PostgreSQL \u002F SQLite",[38,3204,3205],{},"通过 MCP 调 GitHub \u002F Slack \u002F Linear",[38,3207,3208],{},"通过 MCP 调自家内部系统（写一个 MCP Server 即可）",[1674,3210,3212],{"id":3211},"api-first","API-first",[25,3214,3215],{},"每个 app 自动暴露 REST API，参数和返回结构自动生成 OpenAPI Schema。集成到自家产品里不需要写包装代码，给前端 \u002F 微信小程序 \u002F 飞书机器人调用都方便。",[20,3217,1932],{"id":1932},[1674,3219,3221],{"id":3220},"云版difyai","云版（dify.ai）",[25,3223,3224,3225,3230],{},"根据 ",[530,3226,3229],{"href":3227,"rel":3228},"https:\u002F\u002Fwww.tooljunction.io\u002Fai-tools\u002Fdify-ai",[559],"tooljunction.io 2026 评测"," 引用的官方定价：",[97,3232,3233,3245],{},[100,3234,3235],{},[103,3236,3237,3240,3242],{},[106,3238,3239],{},"套餐",[106,3241,95],{},[106,3243,3244],{},"主要限制",[115,3246,3247,3258,3269,3280],{},[103,3248,3249,3252,3255],{},[120,3250,3251],{},"Sandbox",[120,3253,3254],{},"免费",[120,3256,3257],{},"200 次模型调用，1 app，5MB 知识库",[103,3259,3260,3263,3266],{},[120,3261,3262],{},"Professional",[120,3264,3265],{},"$59\u002F月起",[120,3267,3268],{},"5000 调用\u002F月，多 app，50MB 知识库",[103,3270,3271,3274,3277],{},[120,3272,3273],{},"Team",[120,3275,3276],{},"$159\u002F月起",[120,3278,3279],{},"团队协作、SSO",[103,3281,3282,3284,3286],{},[120,3283,155],{},[120,3285,158],{},[120,3287,3288],{},"定制 SLA、私有云",[25,3290,3291,3292,3295],{},"注意：云版价格只是 Dify 平台费，",[41,3293,3294],{},"模型 API 费用另算","（自带 OpenAI \u002F Anthropic key）。",[1674,3297,3299],{"id":3298},"自托管推荐","自托管（推荐）",[25,3301,3302,3307],{},[530,3303,3306],{"href":3304,"rel":3305},"https:\u002F\u002Fgithub.com\u002Flanggenius\u002Fdify",[559],"官方 GitHub 仓库"," 提供 Docker Compose 部署，社区版完全免费可商用：",[3309,3310,3314],"pre",{"className":3311,"code":3312,"language":3313,"meta":569,"style":569},"language-bash shiki shiki-themes github-light github-dark","git clone https:\u002F\u002Fgithub.com\u002Flanggenius\u002Fdify.git\ncd dify\u002Fdocker\ncp .env.example .env\ndocker compose up -d\n# 默认 http:\u002F\u002Flocalhost \u002F 端口可在 .env 调整\n","bash",[195,3315,3316,3332,3341,3352,3365],{"__ignoreMap":569},[3317,3318,3321,3325,3329],"span",{"class":3319,"line":3320},"line",1,[3317,3322,3324],{"class":3323},"sScJk","git",[3317,3326,3328],{"class":3327},"sZZnC"," clone",[3317,3330,3331],{"class":3327}," https:\u002F\u002Fgithub.com\u002Flanggenius\u002Fdify.git\n",[3317,3333,3334,3338],{"class":3319,"line":573},[3317,3335,3337],{"class":3336},"sj4cs","cd",[3317,3339,3340],{"class":3327}," dify\u002Fdocker\n",[3317,3342,3343,3346,3349],{"class":3319,"line":570},[3317,3344,3345],{"class":3323},"cp",[3317,3347,3348],{"class":3327}," .env.example",[3317,3350,3351],{"class":3327}," .env\n",[3317,3353,3354,3356,3359,3362],{"class":3319,"line":602},[3317,3355,598],{"class":3323},[3317,3357,3358],{"class":3327}," compose",[3317,3360,3361],{"class":3327}," up",[3317,3363,3364],{"class":3336}," -d\n",[3317,3366,3367],{"class":3319,"line":603},[3317,3368,3370],{"class":3369},"sJ8bj","# 默认 http:\u002F\u002Flocalhost \u002F 端口可在 .env 调整\n",[25,3372,3373],{},"硬件门槛（社区共识，非官方硬性要求）：",[35,3375,3376,3382,3388],{},[38,3377,3378,3381],{},[41,3379,3380],{},"最低","：2 核 4G，纯外接 API 模式",[38,3383,3384,3387],{},[41,3385,3386],{},"推荐","：4 核 8G + 至少 30GB 磁盘（向量数据 + 文件存储）",[38,3389,3390,3393],{},[41,3391,3392],{},"企业","：8 核 16G+，单机日活上千",[1674,3395,3397],{"id":3396},"真实-tco","真实 TCO",[25,3399,3400],{},"按一家中小团队 3 年场景估算（基于上面引用的多份评测交叉对比）：",[97,3402,3403,3416],{},[100,3404,3405],{},[103,3406,3407,3410,3413],{},[106,3408,3409],{},"成本项",[106,3411,3412],{},"云版 Professional",[106,3414,3415],{},"自托管",[115,3417,3418,3428,3438,3449],{},[103,3419,3420,3423,3426],{},[120,3421,3422],{},"平台费",[120,3424,3425],{},"~$2,100（3 年）",[120,3427,125],{},[103,3429,3430,3433,3435],{},[120,3431,3432],{},"服务器",[120,3434,125],{},[120,3436,3437],{},"~$50\u002F月 × 36 = $1,800",[103,3439,3440,3443,3446],{},[120,3441,3442],{},"模型 API",[120,3444,3445],{},"与下同",[120,3447,3448],{},"与上同",[103,3450,3451,3454,3457],{},[120,3452,3453],{},"运维人力",[120,3455,3456],{},"0",[120,3458,3459],{},"约 0.2 人月",[25,3461,3462,3465],{},[41,3463,3464],{},"结论","：日活 \u003C 100 用云版省心；> 500 或数据敏感场景自托管 ROI 更好。",[20,3467,1968],{"id":1967},[3309,3469,3471],{"className":3311,"code":3470,"language":3313,"meta":569,"style":569},"# 1. 自托管（社区版）\ngit clone https:\u002F\u002Fgithub.com\u002Flanggenius\u002Fdify.git\ncd dify\u002Fdocker\ncp .env.example .env\ndocker compose up -d\n\n# 2. 浏览器打开 http:\u002F\u002Flocalhost\n#    首次会让你创建 admin 账号\n\n# 3. 进入\"设置 → 模型供应商\"，配置 OpenAI \u002F 国产模型 API key\n\n# 4. 在主界面\"创建空白应用\"，选 Chatflow 或 Workflow\n# 5. 拖入\"开始 → LLM → 结束\"节点试一下基础 prompt\n# 6. 满意了点右上\"发布\"，自动生成 API endpoint\n",[195,3472,3473,3478,3486,3492,3500,3510,3516,3522,3528,3533,3539,3544,3550,3556],{"__ignoreMap":569},[3317,3474,3475],{"class":3319,"line":3320},[3317,3476,3477],{"class":3369},"# 1. 自托管（社区版）\n",[3317,3479,3480,3482,3484],{"class":3319,"line":573},[3317,3481,3324],{"class":3323},[3317,3483,3328],{"class":3327},[3317,3485,3331],{"class":3327},[3317,3487,3488,3490],{"class":3319,"line":570},[3317,3489,3337],{"class":3336},[3317,3491,3340],{"class":3327},[3317,3493,3494,3496,3498],{"class":3319,"line":602},[3317,3495,3345],{"class":3323},[3317,3497,3348],{"class":3327},[3317,3499,3351],{"class":3327},[3317,3501,3502,3504,3506,3508],{"class":3319,"line":603},[3317,3503,598],{"class":3323},[3317,3505,3358],{"class":3327},[3317,3507,3361],{"class":3327},[3317,3509,3364],{"class":3336},[3317,3511,3513],{"class":3319,"line":3512},6,[3317,3514,3515],{"emptyLinePlaceholder":593},"\n",[3317,3517,3519],{"class":3319,"line":3518},7,[3317,3520,3521],{"class":3369},"# 2. 浏览器打开 http:\u002F\u002Flocalhost\n",[3317,3523,3525],{"class":3319,"line":3524},8,[3317,3526,3527],{"class":3369},"#    首次会让你创建 admin 账号\n",[3317,3529,3531],{"class":3319,"line":3530},9,[3317,3532,3515],{"emptyLinePlaceholder":593},[3317,3534,3536],{"class":3319,"line":3535},10,[3317,3537,3538],{"class":3369},"# 3. 进入\"设置 → 模型供应商\"，配置 OpenAI \u002F 国产模型 API key\n",[3317,3540,3542],{"class":3319,"line":3541},11,[3317,3543,3515],{"emptyLinePlaceholder":593},[3317,3545,3547],{"class":3319,"line":3546},12,[3317,3548,3549],{"class":3369},"# 4. 在主界面\"创建空白应用\"，选 Chatflow 或 Workflow\n",[3317,3551,3553],{"class":3319,"line":3552},13,[3317,3554,3555],{"class":3369},"# 5. 拖入\"开始 → LLM → 结束\"节点试一下基础 prompt\n",[3317,3557,3559],{"class":3319,"line":3558},14,[3317,3560,3561],{"class":3369},"# 6. 满意了点右上\"发布\"，自动生成 API endpoint\n",[20,3563,3564],{"id":3564},"国内使用注意事项",[237,3566,3567,3573,3579,3585],{},[38,3568,3569,3572],{},[41,3570,3571],{},"云版 dify.ai 直连国内访问稳定但需要付款","——支持国际信用卡 \u002F Stripe",[38,3574,3575,3578],{},[41,3576,3577],{},"自托管 + 国产模型"," = 完全国内闭环，是 Dify 在国内最大优势",[38,3580,3581,3584],{},[41,3582,3583],{},"Docker 镜像拉取","：国内可能慢，建议配 Docker registry 镜像（阿里云 \u002F 网易）",[38,3586,3587,3590,3591,3593],{},[41,3588,3589],{},"数据合规","：完全自托管时，数据零外泄；某些金融 \u002F 政府客户因此从 ",[530,3592,1616],{"href":3045}," 迁到 Dify",[20,3595,2010],{"id":2010},[97,3597,3598,3618],{},[100,3599,3600],{},[103,3601,3602,3604,3606,3610,3614],{},[106,3603,276],{},[106,3605,284],{},[106,3607,3608],{},[530,3609,1616],{"href":3045},[106,3611,3612],{},[530,3613,839],{"href":2030},[106,3615,3616],{},[530,3617,287],{"href":2345},[115,3619,3620,3633,3645,3658,3670,3683,3697,3710],{},[103,3621,3622,3624,3626,3628,3630],{},[120,3623,2043],{},[120,3625,347],{},[120,3627,872],{},[120,3629,347],{},[120,3631,3632],{},"✅（fair-code）",[103,3634,3635,3637,3639,3641,3643],{},[120,3636,2056],{},[120,3638,347],{},[120,3640,872],{},[120,3642,347],{},[120,3644,347],{},[103,3646,3647,3649,3651,3654,3656],{},[120,3648,1334],{},[120,3650,2095],{},[120,3652,3653],{},"★★☆☆☆ 最简单",[120,3655,2095],{},[120,3657,2089],{},[103,3659,3660,3662,3664,3666,3668],{},[120,3661,2086],{},[120,3663,2092],{},[120,3665,2089],{},[120,3667,2095],{},[120,3669,2092],{},[103,3671,3672,3674,3676,3678,3680],{},[120,3673,877],{},[120,3675,2089],{},[120,3677,2095],{},[120,3679,2092],{},[120,3681,3682],{},"★★☆☆☆",[103,3684,3685,3688,3690,3692,3695],{},[120,3686,3687],{},"模型生态",[120,3689,2092],{},[120,3691,2089],{},[120,3693,3694],{},"★★★☆☆（OneAPI 中转）",[120,3696,2089],{},[103,3698,3699,3702,3704,3706,3708],{},[120,3700,3701],{},"中文场景",[120,3703,2089],{},[120,3705,2092],{},[120,3707,2089],{},[120,3709,2095],{},[103,3711,3712,3715,3717,3720,3722],{},[120,3713,3714],{},"字节生态绑定",[120,3716,872],{},[120,3718,3719],{},"✅（飞书\u002F抖音深度集成）",[120,3721,872],{},[120,3723,872],{},[25,3725,3726,3728,3729,2162,3732,3735],{},[41,3727,2155],{},"（基于 ",[530,3730,2161],{"href":2159,"rel":3731},[559],[530,3733,2166],{"href":1661,"rel":3734},[559]," 两份选型指南综合）：",[35,3737,3738,3744,3751,3757,3764],{},[38,3739,3740,3743],{},[41,3741,3742],{},"数据必须不出内网 + 工作流复杂"," → Dify",[38,3745,3746,2180,3749],{},[41,3747,3748],{},"个人 \u002F 小团队 \u002F 快速原型 + 字节生态",[530,3750,1616],{"href":3045},[38,3752,3753,2180,3755],{},[41,3754,2187],{},[530,3756,839],{"href":2030},[38,3758,3759,2180,3762],{},[41,3760,3761],{},"重点是连接外部 SaaS（Slack \u002F Notion \u002F 数据库）",[530,3763,287],{"href":2345},[38,3765,3766,2180,3769],{},[41,3767,3768],{},"要画图式表达 LangChain pipeline",[530,3770,281],{"href":2326},[20,3772,2200],{"id":2200},[35,3774,3775,3781,3797,3808,3821,3827,3833,3839],{},[38,3776,3777,3780],{},[41,3778,3779],{},"社区版与企业版差距比想象大","：多路召回 \u002F 重排序 \u002F 单点登录 \u002F 审计日志都在企业版。社区版做生产前心里要有数。",[38,3782,3783,2892,3789,3792,3793,3796],{},[41,3784,3785,3788],{},[195,3786,3787],{},".env"," 文件改完忘 restart",[195,3790,3791],{},"docker compose down && up -d","，不是 ",[195,3794,3795],{},"restart","——后者不重新加载 env。",[38,3798,3799,2892,3802,3807],{},[41,3800,3801],{},"大版本升级会破坏数据库 schema",[530,3803,3806],{"href":3804,"rel":3805},"https:\u002F\u002Fdocs.dify.ai\u002Fzh-hans",[559],"官方升级文档"," 有详细 migration 步骤，跨大版本（如 0.x → 1.x）务必先备份 PostgreSQL 卷。生产环境强烈建议跑 staging 完整验证后再升。",[38,3809,3810,3813,3814,3816,3817,3820],{},[41,3811,3812],{},"RAG 文件大小社区版默认 15MB","：根据上述知乎实测，超过会失败。改 ",[195,3815,3787],{}," 的 ",[195,3818,3819],{},"UPLOAD_FILE_SIZE_LIMIT"," 并重启容器。",[38,3822,3823,3826],{},[41,3824,3825],{},"代码节点的 Sandbox 性能差","：内置代码执行节点跑在隔离容器里启动慢、内存小。生产高频用建议改成 HTTP 节点调外部服务。",[38,3828,3829,3832],{},[41,3830,3831],{},"工作流\"迭代节点\"循环上限","：默认 10 次，复杂 ReAct agent 容易撞天花板，需要在节点设置里调高。",[38,3834,3835,3838],{},[41,3836,3837],{},"Dify Plugin 系统是新东西","：1.0 后引入的 Plugin 体系替代了原来的 Tools\u002FModels 配置方式，老教程可能已过时——以最新官方文档为准。",[38,3840,3841,3844],{},[41,3842,3843],{},"国内 Docker 拉取镜像慢","：先配国内 registry，否则首次 pull 可能要 30+ 分钟。",[20,3846,459],{"id":458},[25,3848,2278],{},[35,3850,3851,3854,3857,3860,3863,3866],{},[38,3852,3853],{},"中大型企业 LLM 中台建设",[38,3855,3856],{},"需要私有化部署（金融 \u002F 医疗 \u002F 政府）",[38,3858,3859],{},"想做\"AI 工作流即产品\"的开发团队",[38,3861,3862],{},"同时需要 RAG + Agent + Workflow 三件套",[38,3864,3865],{},"想用国产模型 + 国际模型混合编排",[38,3867,3868],{},"已经接受 Docker + 一定运维投入",[25,3870,2298],{},[35,3872,3873,3879,3885,3888,3894],{},[38,3874,3875,3876,3878],{},"纯个人玩家做对话机器人（",[530,3877,1616],{"href":3045}," 更快）",[38,3880,3881,3882,3884],{},"只想做企业知识库 QA（",[530,3883,839],{"href":2030}," RAG 更专）",[38,3886,3887],{},"团队完全没运维能力（云版还行，自托管会踩坑）",[38,3889,3890,3891,3893],{},"需要深度对接字节飞书 \u002F 抖音（",[530,3892,1616],{"href":3045}," 原生）",[38,3895,3896,3897,3899],{},"工作流核心是连接 100+ SaaS（",[530,3898,287],{"href":2345}," 节点更全）",[20,3901,526],{"id":526},[35,3903,3904,3914,3926,3941],{},[38,3905,2334,3906,2258,3908,2258,3910,2258,3912],{},[530,3907,1616],{"href":3045},[530,3909,839],{"href":2030},[530,3911,287],{"href":2345},[530,3913,281],{"href":2326},[38,3915,3916,3917,2258,3919,2258,3921,2258,3924],{},"概念基础：",[530,3918,2352],{"href":2351},[530,3920,341],{"href":2355},[530,3922,3923],{"href":3195},"MCP",[530,3925,2359],{"href":2358},[38,3927,3928,3929,2258,3933,2258,3937,2258,3939],{},"模型选型：",[530,3930,3932],{"href":3931},"\u002Fmodels\u002Fgpt-5.html","GPT-5",[530,3934,3936],{"href":3935},"\u002Fmodels\u002Fclaude-sonnet-4.html","Claude Sonnet 4",[530,3938,2373],{"href":2372},[530,3940,2381],{"href":2380},[38,3942,2384,3943,2258,3947],{},[530,3944,3946],{"href":3945},"\u002Fwiki\u002Ffine-tuning-vs-rag.html","Fine-tuning vs RAG",[530,3948,2392],{"href":2391},[20,3950,545],{"id":545},[35,3952,3953,3960,3966,3972,3979],{},[38,3954,3955,3956],{},"官网：",[530,3957,3958],{"href":3958,"rel":3959},"https:\u002F\u002Fdify.ai",[559],[38,3961,3962,3963],{},"中文文档：",[530,3964,3804],{"href":3804,"rel":3965},[559],[38,3967,3968,3969],{},"GitHub：",[530,3970,3304],{"href":3304,"rel":3971},[559],[38,3973,3974,3975],{},"官方定价：",[530,3976,3977],{"href":3977,"rel":3978},"https:\u002F\u002Fdify.ai\u002Fpricing",[559],[38,3980,3981],{},"第三方评测：tooljunction.io \u002F chatforest.com \u002F besthub.dev \u002F joshuaopolko.com \u002F 知乎 LLM 实战笔记",[25,3983,3984,3985,2429],{},"本卡片由 AIHO 编辑部根据官方公开资料与第三方评测整理。所有事实点均标注来源；如发现版本号 \u002F 价格 \u002F 功能与最新官方信息不一致，请通过 ",[530,3986,2428],{"href":2427},[3988,3989,3990],"style",{},"html pre.shiki code .sScJk, html code.shiki .sScJk{--shiki-default:#6F42C1;--shiki-dark:#B392F0}html pre.shiki code .sZZnC, html code.shiki .sZZnC{--shiki-default:#032F62;--shiki-dark:#9ECBFF}html pre.shiki code .sj4cs, html code.shiki .sj4cs{--shiki-default:#005CC5;--shiki-dark:#79B8FF}html pre.shiki code .sJ8bj, html code.shiki .sJ8bj{--shiki-default:#6A737D;--shiki-dark:#6A737D}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"title":569,"searchDepth":570,"depth":570,"links":3992},[3993,3994,4001,4006,4007,4008,4009,4010,4011,4012],{"id":22,"depth":573,"text":23},{"id":1672,"depth":573,"text":1672,"children":3995},[3996,3997,3998,3999,4000],{"id":3056,"depth":570,"text":3057},{"id":3133,"depth":570,"text":3134},{"id":3170,"depth":570,"text":3171},{"id":3188,"depth":570,"text":3189},{"id":3211,"depth":570,"text":3212},{"id":1932,"depth":573,"text":1932,"children":4002},[4003,4004,4005],{"id":3220,"depth":570,"text":3221},{"id":3298,"depth":570,"text":3299},{"id":3396,"depth":570,"text":3397},{"id":1967,"depth":573,"text":1968},{"id":3564,"depth":573,"text":3564},{"id":2010,"depth":573,"text":2010},{"id":2200,"depth":573,"text":2200},{"id":458,"depth":573,"text":459},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},"\u002Fimg\u002Ftools\u002Fdify.webp","Dify 2026 真实评测：开源 LLMOps 与 AI Agent 平台，集工作流编排、RAG 知识库、Agent、MCP 和多模型接入于一体。本文对比 Coze、FastGPT、n8n，整理自托管部署、云版价格、适合团队和避坑建议。",[2449,590,4016],"ja",{},"\u002Ftools\u002Fagent\u002Fplatform\u002Fdify",[1098,1099,597,598],[4021,4025,4029,4033],{"plan":4022,"price":125,"features":4023,"notes":4024},"Self-hosted（开源版）","Docker 一键部署 + 全部核心功能（工作流 \u002F RAG \u002F Agent \u002F MCP）+ 接任意模型 API","私有部署 \u002F 完全免费 \u002F Apache 2.0",{"plan":4026,"price":125,"features":4027,"notes":4028},"Cloud Sandbox（免费云）","官方托管试水档，含基础调用配额","免运维 \u002F 试水 POC",{"plan":4030,"price":3265,"features":4031,"notes":4032},"Cloud Professional","更高调用额度 + 团队协作 + 商用支持","商用云首选",{"plan":4034,"price":4035,"features":4036,"notes":158},"Cloud Team \u002F Enterprise","Custom","更大配额 + SLA + 私有部署支持 + 合规","云版 SaaS（免费档 \u002F Professional $59\u002F月起） + 开源自托管完全免费",[2488,2489,2490,2491],{"power":603,"ux":602,"price":603,"cn_support":602,"stability":602},{"title":284,"description":4014},"Dify 评测 2026：开源 LLMOps 与 AI Agent 平台，自托管指南",[4043,4045,4047,4049,4051],{"title":4044,"url":3804},"Dify 官方文档（中文）",{"title":4046,"url":3304},"Dify GitHub",{"title":4048,"url":3977},"Dify 官方定价",{"title":4050,"url":2159},"Coze vs Dify vs FastGPT 选型",{"title":4052,"url":4053},"Dify Self-Hosted Guide 2026","https:\u002F\u002Fjoshuaopolko.com\u002Fdify-self-hosted-guide","tools\u002Fagent\u002Fplatform\u002Fdify","开源 LLMOps 平台，私有部署 Agent 首选",[613,614,1111,1112,2517,4057,4058],"llmops","mcp","想私有部署、想接全球任意模型，Dify 是答案。比 Coze 工程化、上手陡一点；比 FastGPT 工作流强、RAG 略弱。","p5aiXfjt5rD0m3qxj903DVwZMIONVWKdagLa7niYhcE",{"id":4062,"title":839,"alternatives":4063,"api_compatible":4064,"body":4066,"category":584,"chinese_friendly":603,"cover":5067,"description":5068,"domestic":587,"extension":588,"faq":15,"free":587,"github":5030,"languages":5069,"lastVerified":15,"meta":5070,"models":5071,"navigation":593,"notSuitable":5075,"opensource":593,"path":5079,"pillar":595,"platforms":5080,"priceTable":5081,"pricing":5103,"published":2486,"relatedPlaybooks":5104,"relatedReviews":5106,"score":5107,"self_host":593,"seo":5108,"seoTitle":5109,"slug":625,"sources":5110,"stem":5121,"suitable":5122,"tagline":5128,"tags":5129,"updated":2520,"verdict":5133,"website":5024,"__hash__":5134},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Ffastgpt.md",[13,2495,12,14],[4065],"openai",{"type":17,"value":4067,"toc":5049},[4068,4070,4099,4110,4112,4116,4119,4133,4136,4162,4166,4173,4205,4212,4216,4269,4276,4279,4282,4285,4292,4338,4344,4348,4379,4383,4489,4492,4495,4499,4507,4628,4643,4645,4799,4807,4841,4847,4849,4923,4925,4927,4947,4949,4967,4969,5016,5018,5041,5046],[20,4069,23],{"id":22},[1625,4071,4073,4088],{"className":4072},[1628,1629,1630],[25,4074,4075,4077,4078,4083,4084,4087],{},[41,4076,1635],{}," labring 团队开源的 LLM 知识库 RAG 平台，27k+ GitHub star（截至 2026-03 数据，",[530,4079,4082],{"href":4080,"rel":4081},"https:\u002F\u002Fcloud.tencent.com\u002Fdeveloper\u002Farticle\u002F2632669",[559],"腾讯云 2026-03 教程"," 引用），Apache 2.0 许可证可商用。",[41,4085,4086],{},"核心优势是 RAG 链路工程做得极细","——问题预处理、混合检索、重排序、上下文组装、答案生成每一步都可视化调参。",[25,4089,4090,4091,4094,4095,4098],{},"最大价值在 ",[41,4092,4093],{},"国内企业知识库 + 私有部署"," 场景。代价是 ",[41,4096,4097],{},"配置门槛","：docker 基础 + 网络知识 + 一定运维能力。",[163,4100,4101],{},[25,4102,4103,4104,4109],{},"来源说明：本文基于 fastgpt.io 官方页面、github.com\u002Flabring\u002FFastGPT 仓库、",[530,4105,4108],{"href":4106,"rel":4107},"https:\u002F\u002Fwww.nanhuantech.com\u002Fzh\u002Fai-reviews\u002Ffastgpt-2025-review",[559],"南环 AI 2026-05 评测","、腾讯云开发者社区 2026-03 部署教程综合整理。版本迭代较快，命令和价格请以最新官方文档为准。",[20,4111,1672],{"id":1672},[1674,4113,4115],{"id":4114},"知识库管理核心能力","知识库管理（核心能力）",[25,4117,4118],{},"支持文件类型：",[35,4120,4121,4124,4127,4130],{},[38,4122,4123],{},"文档：PDF \u002F Word \u002F Markdown \u002F TXT \u002F HTML",[38,4125,4126],{},"表格：Excel \u002F CSV",[38,4128,4129],{},"网页：URL 抓取 + 定时同步",[38,4131,4132],{},"API：通过接口推送内容",[25,4134,4135],{},"处理流程：上传 → 文本切分 → 向量化 → 存储 → 可用于问答。支持：",[35,4137,4138,4144,4150,4156],{},[38,4139,4140,4143],{},[41,4141,4142],{},"文件夹分组","：不同主题 \u002F 部门分类",[38,4145,4146,4149],{},[41,4147,4148],{},"多种分块策略","：默认按段落 \u002F 按 token 数 \u002F 自定义",[38,4151,4152,4155],{},[41,4153,4154],{},"批量导入","：脚本化大批量同步",[38,4157,4158,4161],{},[41,4159,4160],{},"定时同步","：网页源自动更新",[1674,4163,4165],{"id":4164},"rag-流程编排最强卖点","RAG 流程编排（最强卖点）",[25,4167,4168,4172],{},[530,4169,4171],{"href":4106,"rel":4170},[559],"南环 AI 2026 评测"," 总结的 FastGPT RAG 链路：",[237,4174,4175,4181,4187,4193,4199],{},[38,4176,4177,4180],{},[41,4178,4179],{},"问题预处理","：改写 \u002F 扩展 \u002F 错词纠正（提升召回率）",[38,4182,4183,4186],{},[41,4184,4185],{},"检索策略","：语义检索 \u002F 关键词 BM25 \u002F 混合检索，可调相似度阈值",[38,4188,4189,4192],{},[41,4190,4191],{},"重排序（Rerank）","：对初步检索结果二次排序，提升相关性",[38,4194,4195,4198],{},[41,4196,4197],{},"上下文组装","：最优 chunk + 问题 → prompt",[38,4200,4201,4204],{},[41,4202,4203],{},"答案生成","：调大模型基于检索结果回答 + 引用标注",[25,4206,4207,4208,4211],{},"每一步都可视化调参，这是 FastGPT 比 Coze \u002F Dify 在 ",[41,4209,4210],{},"纯知识库 QA 精度","上更高的原因。",[1674,4213,4215],{"id":4214},"多模型支持不绑定厂商","多模型支持（不绑定厂商）",[97,4217,4218,4228],{},[100,4219,4220],{},[103,4221,4222,4225],{},[106,4223,4224],{},"模型类别",[106,4226,4227],{},"支持",[115,4229,4230,4238,4245,4253,4261],{},[103,4231,4232,4235],{},[120,4233,4234],{},"国产闭源",[120,4236,4237],{},"豆包 \u002F 通义千问 \u002F 文心一言 \u002F 智谱 GLM \u002F Moonshot Kimi \u002F MiniMax",[103,4239,4240,4242],{},[120,4241,2043],{},[120,4243,4244],{},"LLaMA \u002F Qwen \u002F ChatGLM \u002F DeepSeek 等可自部署",[103,4246,4247,4250],{},[120,4248,4249],{},"OpenAI 系",[120,4251,4252],{},"GPT-5 \u002F GPT-5 mini \u002F o3",[103,4254,4255,4258],{},[120,4256,4257],{},"Claude 系",[120,4259,4260],{},"Sonnet 4 \u002F Opus 4 \u002F Haiku",[103,4262,4263,4266],{},[120,4264,4265],{},"嵌入 \u002F 重排",[120,4267,4268],{},"BGE \u002F m3e \u002F OpenAI text-embedding-3",[25,4270,4271,4272,4275],{},"可以在 ",[41,4273,4274],{},"应用级别","为不同知识库 \u002F 不同场景配置不同模型，做\"低成本 embedding + 高质量 LLM 生成\"组合。",[1674,4277,4278],{"id":4278},"工作流与高级编排",[25,4280,4281],{},"新版本（v4.14.x）支持类似 Dify 的工作流节点编排——条件分支、循环、HTTP 调用、代码节点。能做\"分类 → 路由到不同子知识库 → 不同模型回答\"这类复杂场景。",[1674,4283,4284],{"id":4284},"多向量库选择",[25,4286,4287,4291],{},[530,4288,4290],{"href":4080,"rel":4289},[559],"腾讯云教程"," 公开的 4 种向量后端：",[97,4293,4294,4304],{},[100,4295,4296],{},[103,4297,4298,4301],{},[106,4299,4300],{},"后端",[106,4302,4303],{},"适用",[115,4305,4306,4314,4322,4330],{},[103,4307,4308,4311],{},[120,4309,4310],{},"PgVector",[120,4312,4313],{},"5000 万索引以下，新手 \u002F 小规模",[103,4315,4316,4319],{},[120,4317,4318],{},"Milvus",[120,4320,4321],{},"亿级以上，高性能",[103,4323,4324,4327],{},[120,4325,4326],{},"Zilliz Cloud",[120,4328,4329],{},"Milvus 全托管 SaaS",[103,4331,4332,4335],{},[120,4333,4334],{},"SeekDB \u002F OceanBase",[120,4336,4337],{},"企业级国产化",[25,4339,4340,4341,1929],{},"部署时选对应 ",[195,4342,4343],{},"docker-compose.{pgvector|milvus|...}.yml",[1674,4345,4347],{"id":4346},"api-与-mcp","API 与 MCP",[35,4349,4350,4356,4362,4373],{},[38,4351,4352,4355],{},[41,4353,4354],{},"对话 API","：流式 \u002F 非流式 HTTP，OpenAI 兼容",[38,4357,4358,4361],{},[41,4359,4360],{},"知识库检索 API","：单独调检索（不走生成）做 hybrid pipeline",[38,4363,4364,4367,4368,4372],{},[41,4365,4366],{},"MCP Server","：3005 端口暴露 MCP SSE 服务，可被 ",[530,4369,4371],{"href":4370},"\u002Fcoding\u002Fcli\u002Fclaude-code.html","Claude Code"," 等客户端直接接入",[38,4374,4375,4378],{},[41,4376,4377],{},"Webhook","：回调通知",[20,4380,4382],{"id":4381},"部署-10-分钟docker","部署 10 分钟（Docker）",[3309,4384,4386],{"className":3311,"code":4385,"language":3313,"meta":569,"style":569},"# 克隆代码\ngit clone https:\u002F\u002Fgithub.com\u002Flabring\u002FFastGPT.git\ncd FastGPT\n\n# 切到最新稳定版（参考 GitHub releases）\ngit switch -c 4.14.7.2\n\n# 选向量库版本（个人 \u002F 小规模选 pg）\ncd deploy\u002Fdocker\u002Fcn\nwget https:\u002F\u002Fdoc.fastgpt.cn\u002Fdeploy\u002Fconfig\u002Fconfig.json\n\n# 启动\ndocker-compose -f docker-compose.pg.yml up -d\n\n# 访问 http:\u002F\u002F\u003Cip>:3000，默认账号 root \u002F 1234\n",[195,4387,4388,4393,4402,4409,4413,4418,4431,4435,4440,4447,4455,4459,4464,4479,4483],{"__ignoreMap":569},[3317,4389,4390],{"class":3319,"line":3320},[3317,4391,4392],{"class":3369},"# 克隆代码\n",[3317,4394,4395,4397,4399],{"class":3319,"line":573},[3317,4396,3324],{"class":3323},[3317,4398,3328],{"class":3327},[3317,4400,4401],{"class":3327}," https:\u002F\u002Fgithub.com\u002Flabring\u002FFastGPT.git\n",[3317,4403,4404,4406],{"class":3319,"line":570},[3317,4405,3337],{"class":3336},[3317,4407,4408],{"class":3327}," FastGPT\n",[3317,4410,4411],{"class":3319,"line":602},[3317,4412,3515],{"emptyLinePlaceholder":593},[3317,4414,4415],{"class":3319,"line":603},[3317,4416,4417],{"class":3369},"# 切到最新稳定版（参考 GitHub releases）\n",[3317,4419,4420,4422,4425,4428],{"class":3319,"line":3512},[3317,4421,3324],{"class":3323},[3317,4423,4424],{"class":3327}," switch",[3317,4426,4427],{"class":3336}," -c",[3317,4429,4430],{"class":3336}," 4.14.7.2\n",[3317,4432,4433],{"class":3319,"line":3518},[3317,4434,3515],{"emptyLinePlaceholder":593},[3317,4436,4437],{"class":3319,"line":3524},[3317,4438,4439],{"class":3369},"# 选向量库版本（个人 \u002F 小规模选 pg）\n",[3317,4441,4442,4444],{"class":3319,"line":3530},[3317,4443,3337],{"class":3336},[3317,4445,4446],{"class":3327}," deploy\u002Fdocker\u002Fcn\n",[3317,4448,4449,4452],{"class":3319,"line":3535},[3317,4450,4451],{"class":3323},"wget",[3317,4453,4454],{"class":3327}," https:\u002F\u002Fdoc.fastgpt.cn\u002Fdeploy\u002Fconfig\u002Fconfig.json\n",[3317,4456,4457],{"class":3319,"line":3541},[3317,4458,3515],{"emptyLinePlaceholder":593},[3317,4460,4461],{"class":3319,"line":3546},[3317,4462,4463],{"class":3369},"# 启动\n",[3317,4465,4466,4469,4472,4475,4477],{"class":3319,"line":3552},[3317,4467,4468],{"class":3323},"docker-compose",[3317,4470,4471],{"class":3336}," -f",[3317,4473,4474],{"class":3327}," docker-compose.pg.yml",[3317,4476,3361],{"class":3327},[3317,4478,3364],{"class":3336},[3317,4480,4481],{"class":3319,"line":3558},[3317,4482,3515],{"emptyLinePlaceholder":593},[3317,4484,4486],{"class":3319,"line":4485},15,[3317,4487,4488],{"class":3369},"# 访问 http:\u002F\u002F\u003Cip>:3000，默认账号 root \u002F 1234\n",[25,4490,4491],{},"最低配置：2C4G + 20GB 硬盘 + Docker 28+ + Docker Compose 2.20+。",[25,4493,4494],{},"进入后台 → 账号 → 模型提供商 → 配置至少 1 个对话模型 + 1 个嵌入模型 → 即可开始建知识库。",[20,4496,4498],{"id":4497},"云版-vs-自托管对比","云版 vs 自托管对比",[25,4500,4501,4506],{},[530,4502,4505],{"href":4503,"rel":4504},"https:\u002F\u002Ffastgpt.io\u002Fzh\u002Fprice",[559],"fastgpt.io 官方定价"," 公开数据：",[97,4508,4509,4534],{},[100,4510,4511],{},[103,4512,4513,4515,4517,4520,4523,4526,4528,4531],{},[106,4514,3239],{},[106,4516,95],{},[106,4518,4519],{},"AI 积分",[106,4521,4522],{},"知识库索引",[106,4524,4525],{},"团队",[106,4527,358],{},[106,4529,4530],{},"知识库",[106,4532,4533],{},"QPM",[115,4535,4536,4560,4584,4608],{},[103,4537,4538,4540,4542,4545,4548,4551,4554,4557],{},[120,4539,3254],{},[120,4541,2467],{},[120,4543,4544],{},"100",[120,4546,4547],{},"600",[120,4549,4550],{},"1",[120,4552,4553],{},"10",[120,4555,4556],{},"3",[120,4558,4559],{},"30",[103,4561,4562,4564,4567,4570,4573,4576,4579,4581],{},[120,4563,894],{},[120,4565,4566],{},"¥99\u002F月",[120,4568,4569],{},"4000",[120,4571,4572],{},"6000",[120,4574,4575],{},"5",[120,4577,4578],{},"50",[120,4580,4559],{},[120,4582,4583],{},"300",[103,4585,4586,4589,4592,4595,4598,4600,4603,4605],{},[120,4587,4588],{},"高级",[120,4590,4591],{},"¥599\u002F月",[120,4593,4594],{},"25000",[120,4596,4597],{},"36000",[120,4599,4578],{},[120,4601,4602],{},"200",[120,4604,4544],{},[120,4606,4607],{},"1500",[103,4609,4610,4613,4615,4618,4620,4622,4624,4626],{},[120,4611,4612],{},"定制",[120,4614,2477],{},[120,4616,4617],{},"弹性",[120,4619,4617],{},[120,4621,4617],{},[120,4623,4617],{},[120,4625,4617],{},[120,4627,4617],{},[25,4629,4630,4633,4634,4637,4638,4642],{},[41,4631,4632],{},"云版适合","：不想运维、量小、要快速上线\n",[41,4635,4636],{},"自托管适合","：量大（10 万+ 日问答）、数据敏感、要深度定制——按 ",[530,4639,4641],{"href":4106,"rel":4640},[559],"南环评测"," 估算：\"日均 10 万次问答的企业场景，商业 SaaS 年费数十万，自建 FastGPT + 开源模型只需数万硬件投入\"",[20,4644,2010],{"id":2010},[97,4646,4647,4667],{},[100,4648,4649],{},[103,4650,4651,4653,4655,4659,4663,4665],{},[106,4652,276],{},[106,4654,839],{},[106,4656,4657],{},[530,4658,284],{"href":2025},[106,4660,4661],{},[530,4662,1616],{"href":3045},[106,4664,542],{},[106,4666,623],{},[115,4668,4669,4689,4705,4720,4737,4752,4768,4783],{},[103,4670,4671,4674,4677,4680,4683,4686],{},[120,4672,4673],{},"核心定位",[120,4675,4676],{},"知识库 QA",[120,4678,4679],{},"综合 LLMOps",[120,4681,4682],{},"Bot + 工作流",[120,4684,4685],{},"文档解析+RAG",[120,4687,4688],{},"桌面级 KB",[103,4690,4691,4693,4696,4698,4700,4702],{},[120,4692,2043],{},[120,4694,4695],{},"✅ Apache 2.0",[120,4697,4695],{},[120,4699,872],{},[120,4701,4695],{},[120,4703,4704],{},"✅ MIT",[103,4706,4707,4709,4712,4714,4716,4718],{},[120,4708,2056],{},[120,4710,4711],{},"★★★★★ docker",[120,4713,2092],{},[120,4715,2059],{},[120,4717,2089],{},[120,4719,2092],{},[103,4721,4722,4725,4728,4730,4732,4735],{},[120,4723,4724],{},"RAG 深度",[120,4726,4727],{},"★★★★★ 最细",[120,4729,2089],{},[120,4731,2095],{},[120,4733,4734],{},"★★★★★ 文档解析最强",[120,4736,2095],{},[103,4738,4739,4742,4744,4746,4748,4750],{},[120,4740,4741],{},"工作流",[120,4743,2089],{},[120,4745,2092],{},[120,4747,2089],{},[120,4749,2095],{},[120,4751,3682],{},[103,4753,4754,4756,4759,4761,4764,4766],{},[120,4755,235],{},[120,4757,4758],{},"★★★☆☆ 需 docker",[120,4760,2089],{},[120,4762,4763],{},"★★★★★ 最简单",[120,4765,2095],{},[120,4767,2089],{},[103,4769,4770,4773,4775,4777,4779,4781],{},[120,4771,4772],{},"中文优化",[120,4774,2092],{},[120,4776,2089],{},[120,4778,2092],{},[120,4780,2089],{},[120,4782,2095],{},[103,4784,4785,4788,4791,4793,4795,4797],{},[120,4786,4787],{},"多平台发布",[120,4789,4790],{},"⚠️ API 为主",[120,4792,2089],{},[120,4794,2092],{},[120,4796,2066],{},[120,4798,2066],{},[25,4800,4801,2156,4803,1703],{},[41,4802,2155],{},[530,4804,4806],{"href":4106,"rel":4805},[559],"南环 AI 评测",[35,4808,4809,4815,4822,4829,4835],{},[38,4810,4811,4814],{},[41,4812,4813],{},"核心需求是 RAG 精度"," → FastGPT",[38,4816,4817,2180,4820],{},[41,4818,4819],{},"需要丰富插件 + 复杂工作流 + 多平台发布",[530,4821,284],{"href":2025},[38,4823,4824,2180,4827],{},[41,4825,4826],{},"零代码、快速发布到飞书 \u002F 微信",[530,4828,1616],{"href":3045},[38,4830,4831,4834],{},[41,4832,4833],{},"文档解析（含 OCR \u002F 表格 \u002F 公式）是瓶颈"," → RAGFlow",[38,4836,4837,4840],{},[41,4838,4839],{},"桌面 \u002F 单机使用"," → AnythingLLM",[25,4842,4843,4846],{},[41,4844,4845],{},"很多企业同时用","：FastGPT 做知识库底座 + Coze 做前端 Bot 发布 \u002F 工作流编排。",[20,4848,2200],{"id":2200},[35,4850,4851,4864,4873,4879,4889,4901,4907,4913],{},[38,4852,4853,2892,4856,4859,4860,4863],{},[41,4854,4855],{},"docker-compose 镜像 tag 不一致",[530,4857,4290],{"href":4080,"rel":4858},[559]," 实测的坑——某些版本编排文件的 image tag 与最新 release 不一致，启动报\"镜像找不到\"，手动改 ",[195,4861,4862],{},"image:"," 行为正确版本即可",[38,4865,4866,4869,4870,4872],{},[41,4867,4868],{},"3000 端口冲突","：默认占用 3000（主服务）\u002F 9000（S3 \u002F MinIO）\u002F 3005（MCP）；改 ",[195,4871,4468],{}," 的 ports 映射端口",[38,4874,4875,4878],{},[41,4876,4877],{},"PostgreSQL pgvector 不够用就换 Milvus","：单库索引超 5000 万时 pgvector 查询性能下降，切 Milvus",[38,4880,4881,4884,4885,4888],{},[41,4882,4883],{},"向量库选错代价大","：先评估索引量再选向量后端，迁移要重新 embedding 整库，按 ",[530,4886,4641],{"href":4106,"rel":4887},[559],"：\"新手 \u002F 小规模 PgVector，中大规模 Milvus，企业 \u002F 国产 OceanBase\"",[38,4890,4891,2892,4894,4897,4898],{},[41,4892,4893],{},"MinIO 默认密码",[195,4895,4896],{},"minioadmin\u002Fminioadmin","，",[41,4899,4900],{},"部署到公网前必须改",[38,4902,4903,4906],{},[41,4904,4905],{},"分段策略影响巨大","：默认分段对长法律 \u002F 医疗文档不友好，需调\"按章节\"或\"自定义\"",[38,4908,4909,4912],{},[41,4910,4911],{},"嵌入模型 ≠ 对话模型","：经常有人只配 GPT-4 没配 embedding 模型，知识库无法索引——必须同时配两类",[38,4914,4915,4918,4919,2233],{},[41,4916,4917],{},"云版 AI 积分会过期","：未用完不能跨月累积（按 ",[530,4920,4922],{"href":4503,"rel":4921},[559],"fastgpt.io 定价 FAQ",[20,4924,459],{"id":458},[25,4926,2278],{},[35,4928,4929,4932,4935,4938,4941,4944],{},[38,4930,4931],{},"企业内部知识库（员工手册 \u002F 制度 \u002F 流程）",[38,4933,4934],{},"产品 FAQ \u002F 用户手册问答",[38,4936,4937],{},"医疗 \u002F 法律 \u002F 金融垂直领域知识系统",[38,4939,4940],{},"数据严格不出网 + Apache 2.0 商用",[38,4942,4943],{},"有 docker 运维基础的技术团队",[38,4945,4946],{},"需要把 RAG 当后端服务的开发者（API 接入业务系统）",[25,4948,2298],{},[35,4950,4951,4956,4961,4964],{},[38,4952,4953,4954,2233],{},"完全非技术用户（去 ",[530,4955,1616],{"href":3045},[38,4957,4958,4959,2233],{},"主要需求是工作流 + 插件集成（去 ",[530,4960,284],{"href":2025},[38,4962,4963],{},"文档解析 \u002F OCR 是首要痛点（RAGFlow）",[38,4965,4966],{},"不想自己运维 + 量很小（FastGPT 云免费版起步即可）",[20,4968,526],{"id":526},[35,4970,4971,4980,4996,5010],{},[38,4972,2334,4973,2258,4975,4977,4978],{},[530,4974,284],{"href":2025},[530,4976,1616],{"href":3045}," \u002F RAGFlow \u002F AnythingLLM \u002F ",[530,4979,287],{"href":2345},[38,4981,2348,4982,2258,4984,2258,4988,2258,4991,2258,4994],{},[530,4983,341],{"href":2355},[530,4985,4987],{"href":4986},"\u002Fwiki\u002Fembedding.html","Embedding",[530,4989,4990],{"href":4986},"Vector Database",[530,4992,4993],{"href":2355},"Reranker",[530,4995,2352],{"href":2351},[38,4997,2365,4998,2258,5000,2258,5002,2258,5004,2258,5008],{},[530,4999,2373],{"href":2372},[530,5001,2377],{"href":2376},[530,5003,2381],{"href":2380},[530,5005,5007],{"href":5006},"\u002Fmodels\u002Fkimi-k2.html","Kimi K2",[530,5009,2369],{"href":2368},[38,5011,2384,5012,2258,5014],{},[530,5013,2392],{"href":2391},[530,5015,2388],{"href":2387},[20,5017,545],{"id":545},[35,5019,5020,5026,5032,5038],{},[38,5021,3955,5022],{},[530,5023,5024],{"href":5024,"rel":5025},"https:\u002F\u002Ffastgpt.io",[559],[38,5027,3968,5028],{},[530,5029,5030],{"href":5030,"rel":5031},"https:\u002F\u002Fgithub.com\u002Flabring\u002FFastGPT",[559],[38,5033,5034,5035],{},"定价：",[530,5036,4503],{"href":4503,"rel":5037},[559],[38,5039,5040],{},"第三方评测：南环 AI \u002F 腾讯云开发者社区 \u002F 飞书 AGI 掘金知识库",[25,5042,5043,5044,2429],{},"本卡片由 AIHO 编辑部根据官方公开资料与第三方评测整理。所有事实点均标注来源；如发现价格 \u002F 命令 \u002F 功能与最新官方信息不一致，请通过 ",[530,5045,2428],{"href":2427},[3988,5047,5048],{},"html pre.shiki code .sJ8bj, html code.shiki .sJ8bj{--shiki-default:#6A737D;--shiki-dark:#6A737D}html pre.shiki code .sScJk, html code.shiki .sScJk{--shiki-default:#6F42C1;--shiki-dark:#B392F0}html pre.shiki code .sZZnC, html code.shiki .sZZnC{--shiki-default:#032F62;--shiki-dark:#9ECBFF}html pre.shiki code .sj4cs, html code.shiki .sj4cs{--shiki-default:#005CC5;--shiki-dark:#79B8FF}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"title":569,"searchDepth":570,"depth":570,"links":5050},[5051,5052,5060,5061,5062,5063,5064,5065,5066],{"id":22,"depth":573,"text":23},{"id":1672,"depth":573,"text":1672,"children":5053},[5054,5055,5056,5057,5058,5059],{"id":4114,"depth":570,"text":4115},{"id":4164,"depth":570,"text":4165},{"id":4214,"depth":570,"text":4215},{"id":4278,"depth":570,"text":4278},{"id":4284,"depth":570,"text":4284},{"id":4346,"depth":570,"text":4347},{"id":4381,"depth":573,"text":4382},{"id":4497,"depth":573,"text":4498},{"id":2010,"depth":573,"text":2010},{"id":2200,"depth":573,"text":2200},{"id":458,"depth":573,"text":459},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},"\u002Fimg\u002Ftools\u002Ffastgpt.webp","FastGPT 真实评测：开源 LLM 知识库 RAG 平台，labring 团队出品，27k+ GitHub star。一键 docker-compose 部署、RAG 流程编排可视化、多向量库支持。AIHO 编辑部基于官方文档与社区资料整理，含与 Dify\u002FCoze 对比、避坑指南。",[2449,590],{},[2453,2454,2452,5072,5073,5074],"gpt-4o","claude-sonnet-4","kimi",[5076,5077,5078],"完全零代码 \u002F 不懂 docker 的用户（去 Coze）","Bot 多平台一键发布场景（Coze 强项）","插件 \u002F 工作流复杂集成（去 Dify）","\u002Ftools\u002Fagent\u002Fplatform\u002Ffastgpt",[1098,1099,597],[5082,5086,5091,5095,5099],{"plan":5083,"price":3254,"limit":5084,"cn_pay":2469,"note":5085},"Self-host 开源","全功能 + 全数据本地","Apache 2.0 可商用",{"plan":5087,"price":5088,"limit":5089,"cn_pay":2469,"note":5090},"云免费版","¥0\u002F月","100 AI 积分 + 600 索引 + 3 知识库","试水",{"plan":5092,"price":4566,"limit":5093,"cn_pay":2474,"note":5094},"云基础版","4000 积分 + 6000 索引 + 50 Agent","中小团队 SaaS",{"plan":5096,"price":4591,"limit":5097,"cn_pay":347,"note":5098},"云高级版","25000 积分 + 36000 索引 + 50 成员 + 200 Agent + 1500 QPM","企业级生产",{"plan":5100,"price":2477,"limit":5101,"cn_pay":347,"note":5102},"云定制版","弹性资源 + 深度技术支持 + 专属客户经理","中大型企业","自托管开源免费 \u002F 云版 ¥0-¥599\u002F月",[5105],"onboarding\u002Ffastgpt-getting-started",[2491,2488,2489,2490],{"power":602,"ux":602,"price":603,"cn_support":603,"stability":602},{"title":839,"description":5068},"FastGPT 评测 2026：开源知识库问答平台，AI 工作流引擎，对比 Dify",[5111,5113,5115,5117,5119],{"title":5112,"url":5024},"FastGPT 官网",{"title":5114,"url":5030},"FastGPT GitHub",{"title":5116,"url":4503},"FastGPT 定价页",{"title":5118,"url":4106},"FastGPT 2025 测评（南环 AI）",{"title":5120,"url":4080},"FastGPT 部署教程（腾讯云）","tools\u002Fagent\u002Fplatform\u002Ffastgpt",[5123,5124,5125,5126,5127],"企业内部知识库（员工手册、规章、流程）","产品文档智能问答（FAQ \u002F 用户手册）","垂直领域知识库（医疗、法律、金融）","数据严格不出网的合规场景","需要精细 RAG 流程编排（重排序、混合检索、阈值调节）","开源知识库问答系统，国内私有部署友好",[613,614,1111,1112,5130,5131,5132],"china","knowledge-base","labring","国内企业知识库私有化首选。RAG 召回工程做得很细，可视化调试好用，docker-compose 一键部署。生态插件不如 Dify 丰富。","NAay3javdz1FV9ZaZVXCpoCFdZlJu5B4E0Y3oSzmYLk",{"id":9,"title":10,"alternatives":5136,"api_compatible":15,"body":5137,"category":584,"chinese_friendly":570,"cover":585,"description":586,"domestic":587,"extension":588,"faq":15,"free":587,"github":565,"languages":5534,"lastVerified":591,"meta":5535,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":594,"pillar":595,"platforms":5536,"priceTable":15,"pricing":599,"published":600,"relatedPlaybooks":15,"relatedReviews":15,"score":5537,"self_host":587,"seo":5538,"seoTitle":605,"slug":606,"sources":5539,"stem":610,"suitable":15,"tagline":611,"tags":5542,"updated":591,"verdict":618,"website":557,"__hash__":619},[12,13,14],{"type":17,"value":5138,"toc":5521},[5139,5141,5143,5145,5147,5185,5187,5233,5237,5239,5245,5261,5265,5281,5283,5301,5303,5417,5419,5449,5451,5473,5475,5479,5483,5489,5493,5495,5503,5505,5509],[20,5140,23],{"id":22},[25,5142,27],{},[25,5144,30],{},[20,5146,33],{"id":33},[35,5148,5149,5153,5157,5161,5165,5169,5173,5177,5181],{},[38,5150,5151,44],{},[41,5152,43],{},[38,5154,5155,50],{},[41,5156,49],{},[38,5158,5159,56],{},[41,5160,55],{},[38,5162,5163,62],{},[41,5164,61],{},[38,5166,5167,68],{},[41,5168,67],{},[38,5170,5171,74],{},[41,5172,73],{},[38,5174,5175,80],{},[41,5176,79],{},[38,5178,5179,86],{},[41,5180,85],{},[38,5182,5183,92],{},[41,5184,91],{},[20,5186,95],{"id":95},[97,5188,5189,5199],{},[100,5190,5191],{},[103,5192,5193,5195,5197],{},[106,5194,108],{},[106,5196,95],{},[106,5198,113],{},[115,5200,5201,5209,5217,5225],{},[103,5202,5203,5205,5207],{},[120,5204,122],{},[120,5206,125],{},[120,5208,128],{},[103,5210,5211,5213,5215],{},[120,5212,133],{},[120,5214,136],{},[120,5216,139],{},[103,5218,5219,5221,5223],{},[120,5220,144],{},[120,5222,147],{},[120,5224,150],{},[103,5226,5227,5229,5231],{},[120,5228,155],{},[120,5230,158],{},[120,5232,161],{},[163,5234,5235],{},[25,5236,167],{},[20,5238,171],{"id":170},[163,5240,5241],{},[25,5242,176,5243],{},[41,5244,179],{},[35,5246,5247,5249,5251,5253,5257,5259],{},[38,5248,184],{},[38,5250,187],{},[38,5252,190],{},[38,5254,193,5255,198],{},[195,5256,197],{},[38,5258,201],{},[38,5260,204],{},[25,5262,5263],{},[41,5264,209],{},[35,5266,5267,5269,5271,5273,5275,5277,5279],{},[38,5268,214],{},[38,5270,217],{},[38,5272,220],{},[38,5274,223],{},[38,5276,226],{},[38,5278,229],{},[38,5280,232],{},[20,5282,235],{"id":235},[237,5284,5285,5289,5293,5295,5297,5299],{},[38,5286,241,5287,245],{},[195,5288,244],{},[38,5290,248,5291,252],{},[195,5292,251],{},[38,5294,255],{},[38,5296,258],{},[38,5298,261],{},[38,5300,264],{},[20,5302,267],{"id":267},[97,5304,5305,5319],{},[100,5306,5307],{},[103,5308,5309,5311,5313,5315,5317],{},[106,5310,276],{},[106,5312,10],{},[106,5314,281],{},[106,5316,284],{},[106,5318,287],{},[115,5320,5321,5333,5345,5357,5369,5381,5393,5405],{},[103,5322,5323,5325,5327,5329,5331],{},[120,5324,294],{},[120,5326,297],{},[120,5328,300],{},[120,5330,303],{},[120,5332,306],{},[103,5334,5335,5337,5339,5341,5343],{},[120,5336,311],{},[120,5338,314],{},[120,5340,314],{},[120,5342,319],{},[120,5344,322],{},[103,5346,5347,5349,5351,5353,5355],{},[120,5348,235],{},[120,5350,329],{},[120,5352,329],{},[120,5354,329],{},[120,5356,336],{},[103,5358,5359,5361,5363,5365,5367],{},[120,5360,341],{},[120,5362,344],{},[120,5364,347],{},[120,5366,350],{},[120,5368,353],{},[103,5370,5371,5373,5375,5377,5379],{},[120,5372,358],{},[120,5374,347],{},[120,5376,347],{},[120,5378,350],{},[120,5380,347],{},[103,5382,5383,5385,5387,5389,5391],{},[120,5384,371],{},[120,5386,347],{},[120,5388,347],{},[120,5390,347],{},[120,5392,347],{},[103,5394,5395,5397,5399,5401,5403],{},[120,5396,384],{},[120,5398,387],{},[120,5400,387],{},[120,5402,387],{},[120,5404,387],{},[103,5406,5407,5409,5411,5413,5415],{},[120,5408,398],{},[120,5410,401],{},[120,5412,401],{},[120,5414,406],{},[120,5416,322],{},[20,5418,411],{"id":411},[35,5420,5421,5425,5429,5433,5437,5441,5445],{},[38,5422,5423,419],{},[41,5424,418],{},[38,5426,5427,425],{},[41,5428,424],{},[38,5430,5431,431],{},[41,5432,430],{},[38,5434,5435,437],{},[41,5436,436],{},[38,5438,5439,443],{},[41,5440,442],{},[38,5442,5443,449],{},[41,5444,448],{},[38,5446,5447,455],{},[41,5448,454],{},[20,5450,459],{"id":458},[35,5452,5453,5455,5457,5459,5461,5463,5465,5467,5469,5471],{},[38,5454,464],{},[38,5456,467],{},[38,5458,470],{},[38,5460,473],{},[38,5462,476],{},[38,5464,479],{},[38,5466,482],{},[38,5468,485],{},[38,5470,488],{},[38,5472,491],{},[20,5474,495],{"id":494},[25,5476,5477,501],{},[41,5478,500],{},[25,5480,5481,507],{},[41,5482,506],{},[25,5484,5485,513,5487,517],{},[41,5486,512],{},[195,5488,516],{},[25,5490,5491,523],{},[41,5492,522],{},[20,5494,526],{"id":526},[25,5496,5497,534,5499,534,5501],{},[530,5498,533],{"href":532},[530,5500,538],{"href":537},[530,5502,542],{"href":541},[20,5504,545],{"id":545},[163,5506,5507],{},[25,5508,550],{},[35,5510,5511,5516],{},[38,5512,5513],{},[530,5514,560],{"href":557,"rel":5515},[559],[38,5517,5518],{},[530,5519,567],{"href":565,"rel":5520},[559],{"title":569,"searchDepth":570,"depth":570,"links":5522},[5523,5524,5525,5526,5527,5528,5529,5530,5531,5532,5533],{"id":22,"depth":573,"text":23},{"id":33,"depth":573,"text":33},{"id":95,"depth":573,"text":95},{"id":170,"depth":573,"text":171},{"id":235,"depth":573,"text":235},{"id":267,"depth":573,"text":267},{"id":411,"depth":573,"text":411},{"id":458,"depth":573,"text":459},{"id":494,"depth":573,"text":495},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},[590],{},[597,598],{"power":570,"ux":602,"price":603,"cn_support":570,"stability":570},{"title":10,"description":586},[5540,5541],{"title":560,"url":557},{"title":567,"url":565},[613,614,615,616,617],{"id":5544,"title":281,"alternatives":5545,"api_compatible":15,"body":5548,"category":584,"chinese_friendly":570,"cover":6106,"description":6107,"domestic":587,"extension":588,"faq":6108,"free":587,"github":15,"languages":6121,"lastVerified":15,"meta":6123,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":6124,"pillar":595,"platforms":6125,"priceTable":6127,"pricing":6140,"published":6141,"relatedPlaybooks":6142,"relatedReviews":15,"score":6144,"self_host":593,"seo":6145,"seoTitle":6146,"slug":12,"sources":6147,"stem":6157,"suitable":15,"tagline":6158,"tags":6159,"updated":2520,"verdict":6162,"website":6150,"__hash__":6163},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Flangflow.md",[14,5546,5547],"agent\u002Fprotocol\u002Fcomposio","agent\u002Fgeneral\u002Fopenmanus",{"type":17,"value":5549,"toc":6094},[5550,5552,5555,5558,5560,5634,5636,5661,5666,5670,5674,5700,5704,5730,5732,5788,5791,5814,5816,5953,5955,6010,6012,6038,6040,6060,6062,6092],[20,5551,23],{"id":22},[25,5553,5554],{},"Langflow 是 2023 开源、2024 被 DataStax 收购的可视化 LangChain 画布。20,000+ GitHub stars、MIT 协议、Python 实现、与 Astra DB 向量库深度集成。差异点：拖拽式画布把 LangChain primitive 映射成节点 + RAG pipeline 原生组件 + 多 agent 工作流 + 节点可下钻到 Python 代码 + 自托管 \u002F Docker \u002F DataStax Astra 云托管 + Pinecone \u002F pgvector \u002F 主流向量库适配 + Visual GUI for building LangChain pipelines。Self-host 免费 \u002F Cloud Free + Paid ~$25\u002F月起。",[25,5556,5557],{},"适合：工程团队要可视化建 LangChain 流 + 不被 SaaS 锁住；合规 \u002F 数据驻留要求自托管；要 Astra DB 一站式 RAG；想用 LangChain 但讨厌纯代码协作。不适合：业务侧 + 非工程师（用 Dify \u002F Coze）；纯无代码偏好（Langflow 仍要懂 LangChain 概念）；轻量场景 + 不需要画布（直接写 LangChain 代码）。",[20,5559,33],{"id":33},[35,5561,5562,5568,5574,5580,5586,5592,5598,5604,5610,5616,5622,5628],{},[38,5563,5564,5567],{},[41,5565,5566],{},"可视化 LangChain 画布","：节点 = primitive，连线 = 数据流",[38,5569,5570,5573],{},[41,5571,5572],{},"RAG pipeline 原生组件","：Chunking \u002F Embedding \u002F VectorStore \u002F Retriever",[38,5575,5576,5579],{},[41,5577,5578],{},"多 agent 工作流","：编排多 agent 协作",[38,5581,5582,5585],{},[41,5583,5584],{},"Python 下钻","：任意节点可写 custom Python",[38,5587,5588,5591],{},[41,5589,5590],{},"向量库适配","：Astra DB \u002F Pinecone \u002F pgvector \u002F Weaviate \u002F Chroma",[38,5593,5594,5597],{},[41,5595,5596],{},"API 部署","：流程一键导出为 REST API",[38,5599,5600,5603],{},[41,5601,5602],{},"Real-time collaboration","：多用户同 project",[38,5605,5606,5609],{},[41,5607,5608],{},"版本控制","：内置 versioning + revert",[38,5611,5612,5615],{},[41,5613,5614],{},"数据可视化","：node output \u002F data flow 可视化调试",[38,5617,5618,5621],{},[41,5619,5620],{},"角色权限","：user auth + RBAC",[38,5623,5624,5627],{},[41,5625,5626],{},"Docker \u002F pip 安装","：5 分钟启动",[38,5629,5630,5633],{},[41,5631,5632],{},"Astra-hosted cloud","：DataStax 托管选项",[20,5635,95],{"id":95},[35,5637,5638,5644,5650,5656],{},[38,5639,5640,5643],{},[41,5641,5642],{},"Self-Host (OSS)","：$0；MIT 完全免费，自付 LLM API + 服务器",[38,5645,5646,5649],{},[41,5647,5648],{},"Cloud Free","：$0；DataStax Astra-hosted 小流量",[38,5651,5652,5655],{},[41,5653,5654],{},"Cloud Paid","：~$25\u002F月起；Astra DB + 更高额度 + 团队协作",[38,5657,5658,5660],{},[41,5659,155],{},"：联系销售；SSO + audit + 私有部署 + SLA",[163,5662,5663],{},[25,5664,5665],{},"自托管最低成本：$5-10\u002F月 VPS + LLM API token。Cloud Paid $25\u002F月适合不想运维的小团队。",[20,5667,5669],{"id":5668},"实测rag-应用-内部工具搭建","实测（RAG 应用 + 内部工具搭建）",[25,5671,5672],{},[41,5673,179],{},[35,5675,5676,5679,5682,5685,5688,5691,5694,5697],{},[38,5677,5678],{},"画布直观，比纯写 LangChain 协作效率高 5x",[38,5680,5681],{},"节点下钻到 Python 让灵活度不被画布限制",[38,5683,5684],{},"Astra DB 集成省了配 vector store 时间",[38,5686,5687],{},"4.4\u002F5 用户评分（Propicked \u002F Tooliverse）",[38,5689,5690],{},"开源 + 自托管 + 数据驻留满足合规",[38,5692,5693],{},"多 agent 编排比裸 LangChain 调试容易",[38,5695,5696],{},"RAG pipeline 模板一键起 demo",[38,5698,5699],{},"与 DataStax 长期支持降低 abandon ware 风险",[25,5701,5702],{},[41,5703,209],{},[35,5705,5706,5709,5712,5715,5718,5721,5724,5727],{},[38,5707,5708],{},"稳定性偶发：复杂流大流量下 node 偶尔失联（SelectHub 反馈）",[38,5710,5711],{},"第三方 API 依赖：external API 失败时错误处理弱",[38,5713,5714],{},"production readiness 不算 mission-critical（要自加 observability）",[38,5716,5717],{},"LangChain 升级偶尔 break 旧 flow",[38,5719,5720],{},"文档对新组件滞后 1-2 月",[38,5722,5723],{},"中文 UI 不完整，业务侧用户上手陡",[38,5725,5726],{},"大型 flow（100+ 节点）画布卡顿",[38,5728,5729],{},"多人协作偶发同步冲突",[20,5731,235],{"id":235},[3309,5733,5735],{"className":3311,"code":5734,"language":3313,"meta":569,"style":569},"# pip 安装\npip install langflow\nlangflow run  # http:\u002F\u002Flocalhost:7860\n\n# 或 Docker\ndocker run -p 7860:7860 langflowai\u002Flangflow:latest\n",[195,5736,5737,5742,5753,5764,5768,5773],{"__ignoreMap":569},[3317,5738,5739],{"class":3319,"line":3320},[3317,5740,5741],{"class":3369},"# pip 安装\n",[3317,5743,5744,5747,5750],{"class":3319,"line":573},[3317,5745,5746],{"class":3323},"pip",[3317,5748,5749],{"class":3327}," install",[3317,5751,5752],{"class":3327}," langflow\n",[3317,5754,5755,5758,5761],{"class":3319,"line":570},[3317,5756,5757],{"class":3323},"langflow",[3317,5759,5760],{"class":3327}," run",[3317,5762,5763],{"class":3369},"  # http:\u002F\u002Flocalhost:7860\n",[3317,5765,5766],{"class":3319,"line":602},[3317,5767,3515],{"emptyLinePlaceholder":593},[3317,5769,5770],{"class":3319,"line":603},[3317,5771,5772],{"class":3369},"# 或 Docker\n",[3317,5774,5775,5777,5779,5782,5785],{"class":3319,"line":3512},[3317,5776,598],{"class":3323},[3317,5778,5760],{"class":3327},[3317,5780,5781],{"class":3336}," -p",[3317,5783,5784],{"class":3327}," 7860:7860",[3317,5786,5787],{"class":3327}," langflowai\u002Flangflow:latest\n",[25,5789,5790],{},"试 RAG 流：",[237,5792,5793,5796,5799,5802,5805,5808,5811],{},[38,5794,5795],{},"新建 flow → 选 Document QA 模板",[38,5797,5798],{},"Document Loader 节点 → 上传 PDF",[38,5800,5801],{},"Splitter → Embedder（OpenAI 或本地）",[38,5803,5804],{},"VectorStore（Astra \u002F Chroma）",[38,5806,5807],{},"Retriever + ChatOpenAI → Chat Output",[38,5809,5810],{},"部署为 API → 拿到 endpoint",[38,5812,5813],{},"复杂场景下钻节点写 Python 自定义",[20,5815,267],{"id":267},[97,5817,5818,5832],{},[100,5819,5820],{},[103,5821,5822,5824,5826,5828,5830],{},[106,5823,276],{},[106,5825,281],{},[106,5827,284],{},[106,5829,287],{},[106,5831,10],{},[115,5833,5834,5851,5865,5879,5892,5908,5921,5938],{},[103,5835,5836,5839,5842,5845,5848],{},[120,5837,5838],{},"中心",[120,5840,5841],{},"LangChain primitive",[120,5843,5844],{},"LLMOps 全平台",[120,5846,5847],{},"通用 workflow",[120,5849,5850],{},"LangChain（JS）",[103,5852,5853,5855,5857,5860,5863],{},[120,5854,2043],{},[120,5856,4704],{},[120,5858,5859],{},"✅ AGPL",[120,5861,5862],{},"✅ Sustainable",[120,5864,4704],{},[103,5866,5867,5869,5872,5875,5877],{},[120,5868,3415],{},[120,5870,5871],{},"✅ pip\u002FDocker",[120,5873,5874],{},"✅ Docker",[120,5876,5874],{},[120,5878,347],{},[103,5880,5881,5883,5886,5888,5890],{},[120,5882,2787],{},[120,5884,5885],{},"✅ 旗舰",[120,5887,347],{},[120,5889,347],{},[120,5891,347],{},[103,5893,5894,5897,5900,5903,5906],{},[120,5895,5896],{},"代码下钻",[120,5898,5899],{},"✅ Python",[120,5901,5902],{},"部分",[120,5904,5905],{},"✅ JS",[120,5907,5905],{},[103,5909,5910,5913,5915,5917,5919],{},[120,5911,5912],{},"RAG 内置",[120,5914,347],{},[120,5916,347],{},[120,5918,5902],{},[120,5920,347],{},[103,5922,5923,5926,5929,5932,5935],{},[120,5924,5925],{},"起价（云）",[120,5927,5928],{},"$25\u002F月",[120,5930,5931],{},"$59\u002F月（Team）",[120,5933,5934],{},"自托管 $0",[120,5936,5937],{},"–",[103,5939,5940,5942,5945,5948,5950],{},[120,5941,398],{},[120,5943,5944],{},"工程 + LangChain",[120,5946,5947],{},"业务 + LLMOps",[120,5949,322],{},[120,5951,5952],{},"JS 生态",[20,5954,411],{"id":411},[35,5956,5957,5963,5969,5975,5981,5987,5993,5999,6004],{},[38,5958,5959,5962],{},[41,5960,5961],{},"自托管推荐 Docker","：pip 版本依赖冲突难调",[38,5964,5965,5968],{},[41,5966,5967],{},"生产加 observability","：langsmith \u002F sentry \u002F prometheus 必装",[38,5970,5971,5974],{},[41,5972,5973],{},"Astra DB cloud free","：起步够用，付费版起步前算清成本",[38,5976,5977,5980],{},[41,5978,5979],{},"复杂 flow 拆模块","：100+ 节点画布卡顿，拆成子 flow",[38,5982,5983,5986],{},[41,5984,5985],{},"LangChain 版本 pin","：Langflow 升级前测试 flow 兼容性",[38,5988,5989,5992],{},[41,5990,5991],{},"第三方 API 加重试","：custom Python 节点写 retry + fallback",[38,5994,5995,5998],{},[41,5996,5997],{},"多人协作 lock","：同时编辑 flow 容易冲突，加 lock 流程",[38,6000,6001,6003],{},[41,6002,3701],{},"：UI 英文为主，业务侧用户先培训",[38,6005,6006,6009],{},[41,6007,6008],{},"Self-host vs Cloud","：合规要 self-host，省心要 Cloud",[20,6011,459],{"id":458},[35,6013,6014,6017,6020,6023,6026,6029,6032,6035],{},[38,6015,6016],{},"✅ 工程团队要可视化建 LangChain 流",[38,6018,6019],{},"✅ 合规 \u002F 数据驻留要求自托管",[38,6021,6022],{},"✅ 要 Astra DB 一站式 RAG",[38,6024,6025],{},"✅ Python 团队 + 想画布 + 想下钻代码",[38,6027,6028],{},"❌ 业务侧 + 非工程师（用 Dify \u002F Coze）",[38,6030,6031],{},"❌ 纯无代码偏好",[38,6033,6034],{},"❌ 轻量场景 + 直接写 LangChain 更快",[38,6036,6037],{},"❌ JS 生态优先（用 Flowise）",[20,6039,526],{"id":526},[35,6041,6042,6048,6054],{},[38,6043,6044],{},[530,6045,6047],{"href":6046},"\u002Ftools\u002Fagent\u002Fplatform\u002Fn8n","n8n 评测",[38,6049,6050],{},[530,6051,6053],{"href":6052},"\u002Ftools\u002Fagent\u002Fprotocol\u002Fcomposio","Composio 评测",[38,6055,6056],{},[530,6057,6059],{"href":6058},"\u002Ftools\u002Fagent\u002Fgeneral\u002Fopenmanus","OpenManus 评测",[20,6061,545],{"id":545},[237,6063,6064,6071,6078,6085],{},[38,6065,6066,6067],{},"Langflow 官网 + 定价 ",[530,6068,6069],{"href":6069,"rel":6070},"https:\u002F\u002Fwww.langflow.org\u002Fpricing",[559],[38,6072,6073,6074],{},"AutomationAtlas — Langflow $25\u002Fmo 2026 Comparison ",[530,6075,6076],{"href":6076,"rel":6077},"https:\u002F\u002Fautomationatlas.io\u002Ftools\u002Flangflow",[559],[38,6079,6080,6081],{},"xpay — LangFlow 2026 Visual AI Agent Builder（MIT + DataStax）",[530,6082,6083],{"href":6083,"rel":6084},"https:\u002F\u002Fwww.xpay.sh\u002Fresources\u002Fagentic-frameworks\u002Flangflow",[559],[38,6086,6087,6088],{},"SelectHub — LangFlow Reviews 2026（稳定性 + 生产 readiness）",[530,6089,6090],{"href":6090,"rel":6091},"https:\u002F\u002Fwww.selecthub.com\u002Fp\u002Fai-agent-builder-software\u002Flangflow\u002F",[559],[3988,6093,5048],{},{"title":569,"searchDepth":570,"depth":570,"links":6095},[6096,6097,6098,6099,6100,6101,6102,6103,6104,6105],{"id":22,"depth":573,"text":23},{"id":33,"depth":573,"text":33},{"id":95,"depth":573,"text":95},{"id":5668,"depth":573,"text":5669},{"id":235,"depth":573,"text":235},{"id":267,"depth":573,"text":267},{"id":411,"depth":573,"text":411},{"id":458,"depth":573,"text":459},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},"\u002Fimg\u002Ftools\u002Flangflow.webp","Langflow 真实评测：2023 开源项目，2024 被 DataStax 收购，与 Astra DB 向量数据库平台深度集成。20,000+ GitHub stars、MIT 协议。差异点：拖拽式可视化画布映射 LangChain 原语 + RAG pipeline 原生组件 + 多 agent 工作流 + Python + 自托管（pip \u002F Docker）+ DataStax Astra 托管云 + Pinecone \u002F pgvector \u002F 主流向量库适配。Cloud 免费档 + 约 $25\u002F月起。",[6109,6112,6115,6118],{"q":6110,"a":6111},"Langflow 和 Dify \u002F n8n \u002F Flowise 怎么选？","Langflow 强『可视化 LangChain 原语 + Python 代码可下钻 + 自托管 + Astra DB 集成』，工程导向。Dify 是 LLMOps 全平台（含 dataset \u002F app \u002F observability），业务侧更友好。n8n 是通用 workflow（非 LangChain 中心），70+ LangChain 节点是后加的。Flowise 是 Langflow 的同类竞品（JS 生态）。工程团队 + Python + 合规自托管 → Langflow；业务 + 完整 LLMOps → Dify；通用自动化 → n8n。",{"q":6113,"a":6114},"为什么被 DataStax 收购？","DataStax 是 Cassandra 商业公司 + Astra DB 向量数据库厂商。收购 Langflow 是为了把『可视化 LangChain builder』和『生产级 vector store』捆成一站式 RAG 解决方案。Langflow 主仓仍是 MIT 开源，但 cloud \u002F 企业版深度依赖 Astra。",{"q":6116,"a":6117},"可视化画布会不会限制灵活度？","Langflow 节点对应 LangChain primitive，可在任意节点下钻到 Python 代码 + 自定义。所以可视化层是『脚手架 + 协作工具』，不是『纯无代码黑盒』。复杂逻辑、custom tool、retriever 都能写代码扩展。",{"q":6119,"a":6120},"production readiness 如何？","用户反馈：原型 + 内部工具非常顺；大流量 \u002F 关键业务要自行加 observability \u002F 错误处理 \u002F 缓存。SelectHub 评测列出『稳定性偶发 + 第三方 API 依赖 + 非完全 production-ready』。生产部署建议 Astra-hosted cloud 或自托管 + 加 sentry \u002F langsmith \u002F prometheus。",[590,6122],"multi",{},"\u002Ftools\u002Fagent\u002Fplatform\u002Flangflow",[1111,6126,598,2463],"cloud",[6128,6131,6133,6137],{"plan":5642,"price":125,"features":6129,"notes":6130},"MIT + pip\u002FDocker + 全部组件 + Pinecone\u002Fpgvector\u002FAstra 适配","自付 LLM API + 服务器",{"plan":5648,"price":125,"features":6132,"notes":5090},"DataStax 托管 + 小流量",{"plan":5654,"price":6134,"features":6135,"notes":6136},"~$25\u002F月起","Astra DB + 更高额度 + 更多并发 + 团队协作","按使用量阶梯",{"plan":155,"price":158,"features":6138,"notes":6139},"SSO + audit + 私有部署 + SLA + 数据驻留","合规 \u002F 大客户","Self-host 完全免费 MIT \u002F Cloud Free + 付费起 ~$25·月（Astra-hosted）","2026-06-19",[6143],"onboarding\u002Frag-app-workflow",{"power":602,"ux":603,"price":603,"cn_support":570,"stability":602},{"title":281,"description":6107},"Langflow 评测 2026：可视化 AI 工作流构建工具，LangChain 低代码平台",[6148,6151,6153,6155],{"name":6149,"url":6150,"accessed":2520},"Langflow 官网","https:\u002F\u002Fwww.langflow.org",{"name":6152,"url":6076,"accessed":2520},"AutomationAtlas — Langflow $25\u002Fmo 2026 Comparison",{"name":6154,"url":6083,"accessed":2520},"xpay — LangFlow 2026 Visual AI Agent Builder",{"name":6156,"url":6090,"accessed":2520},"SelectHub — LangFlow Reviews 2026","tools\u002Fagent\u002Fplatform\u002Flangflow","DataStax 出品的可视化 LangChain 画布——MIT 开源 + 20k+ stars + 自托管 + Astra DB 云托管",[614,6160,617,1112,6161,5757],"visual-builder","datastax","工程团队要可视化建 LangChain 流 + 不被 SaaS 锁住 + 合规要求自托管的最佳选择。要纯无代码 + 业务侧 → Dify \u002F Coze；要纯代码 → 直接写 LangChain \u002F LlamaIndex。","rvh-hO12QKzN5KXHjH_xWXuls1hBO42dYKXttkerPls",{"id":6165,"title":287,"alternatives":6166,"api_compatible":15,"body":6167,"category":584,"chinese_friendly":570,"cover":6813,"description":6814,"domestic":587,"extension":588,"faq":6815,"free":587,"github":15,"languages":6828,"lastVerified":15,"meta":6829,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":6046,"pillar":595,"platforms":6830,"priceTable":6831,"pricing":6847,"published":6141,"relatedPlaybooks":6848,"relatedReviews":15,"score":6850,"self_host":593,"seo":6851,"seoTitle":6852,"slug":14,"sources":6853,"stem":6862,"suitable":15,"tagline":6863,"tags":6864,"updated":2520,"verdict":6866,"website":6775,"__hash__":6867},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fn8n.md",[12,5546,5547],{"type":17,"value":6168,"toc":6801},[6169,6171,6174,6177,6179,6253,6255,6278,6283,6287,6291,6317,6321,6347,6351,6476,6479,6507,6510,6512,6658,6660,6721,6723,6749,6751,6766,6768,6798],[20,6170,23],{"id":22},[25,6172,6173],{},"n8n 是 2019 创立、Sustainable Use License（fair-code）的开源自动化平台。2026 突破 200,000 active users、5x ARR 增长、5,800+ 社区 AI workflow。差异点：近 70 个 LangChain 专属节点 + 原生 MCP 协议 + AI Agent 节点（reasoning loop + 工具调用）+ Ollama \u002F OpenAI 双路 + 400+ 集成 + execution-based 定价（步骤数无关）+ 自托管 VPS $5-10\u002F月跑全套。",[25,6175,6176],{},"适合：开发者 + 想完全控制 + 高量级自动化；从 Zapier \u002F Make 迁出降本；要 AI Agent + LangChain + MCP 一体；合规 \u002F 自托管 \u002F 数据驻留要求。不适合：非技术 + 要 8000+ 现成集成（用 Zapier）；中等复杂 + 不愿自托管（用 Make）；纯研究 \u002F 学术 agent（用 OpenManus \u002F Langflow）。",[20,6178,33],{"id":33},[35,6180,6181,6187,6193,6199,6205,6211,6217,6223,6229,6235,6241,6247],{},[38,6182,6183,6186],{},[41,6184,6185],{},"AI Agent 节点","：reasoning loop + 自主工具选择",[38,6188,6189,6192],{},[41,6190,6191],{},"70 LangChain 节点","：LLM \u002F VectorStore \u002F Agent \u002F Tool \u002F Memory 全套",[38,6194,6195,6198],{},[41,6196,6197],{},"原生 MCP","：MCP server 一键挂载到 agent",[38,6200,6201,6204],{},[41,6202,6203],{},"Ollama 集成","：本地 LLM 零 API 成本",[38,6206,6207,6210],{},[41,6208,6209],{},"400+ 集成","：Slack \u002F GitHub \u002F Google \u002F Notion \u002F 主流 SaaS",[38,6212,6213,6216],{},[41,6214,6215],{},"HTTP \u002F Webhook 万能节点","：任意 REST API 都能接",[38,6218,6219,6222],{},[41,6220,6221],{},"Cron \u002F Trigger","：定时 \u002F 事件 \u002F Webhook 触发",[38,6224,6225,6228],{},[41,6226,6227],{},"多分支并行 + 错误处理","：production workflow 必备",[38,6230,6231,6234],{},[41,6232,6233],{},"版本控制 + Git Sync","：workflow as code",[38,6236,6237,6240],{},[41,6238,6239],{},"自托管 Docker \u002F Kubernetes","：一行起 + 水平扩展",[38,6242,6243,6246],{},[41,6244,6245],{},"execution-based 定价","：20 步和 2 步同价（自托管 = 0）",[38,6248,6249,6252],{},[41,6250,6251],{},"5800+ 社区 workflow","：clone 即用",[20,6254,95],{"id":95},[35,6256,6257,6263,6268,6273],{},[38,6258,6259,6262],{},[41,6260,6261],{},"Community Self-host","：$0；全功能 + 不限 execution + 自付 VPS $5-10\u002F月",[38,6264,6265,6267],{},[41,6266,133],{},"：~€20\u002F月；2,500 executions + 5 workflow",[38,6269,6270,6272],{},[41,6271,144],{},"：~€50\u002F月；高 executions + 团队协作",[38,6274,6275,6277],{},[41,6276,155],{},"：联系销售；SSO + LDAP + 私有部署 + SLA",[163,6279,6280],{},[25,6281,6282],{},"真实场景：Zapier $50\u002F月跑中等复杂 → n8n 自托管 $5\u002F月跑同样的 = 10x 降本。",[20,6284,6286],{"id":6285},"实测中型团队-saas-迁移-本地-ai","实测（中型团队 SaaS 迁移 + 本地 AI）",[25,6288,6289],{},[41,6290,179],{},[35,6292,6293,6296,6299,6302,6305,6308,6311,6314],{},[38,6294,6295],{},"自托管成本几乎可忽略：$5\u002F月 VPS 跑几十个 workflow",[38,6297,6298],{},"AI Agent + Ollama 让 LLM 任务零 API 成本",[38,6300,6301],{},"70 LangChain 节点覆盖 RAG \u002F Agent \u002F 多模态",[38,6303,6304],{},"MCP 原生集成让 n8n agent 调用任意 MCP server",[38,6306,6307],{},"5800+ 社区 workflow 节省 80% 上手时间",[38,6309,6310],{},"HTTP \u002F Webhook 万能节点弥补 native 集成缺口",[38,6312,6313],{},"升级 Docker tag 一行，无 vendor 升级费",[38,6315,6316],{},"中文社区 \u002F B 站教程丰富",[25,6318,6319],{},[41,6320,209],{},[35,6322,6323,6326,6329,6332,6335,6338,6341,6344],{},[38,6324,6325],{},"集成数（400+）远少于 Zapier（8000+），冷门 SaaS 要写 HTTP 自己接",[38,6327,6328],{},"自托管要懂 Docker \u002F Postgres \u002F Redis（高吞吐场景）",[38,6330,6331],{},"Cloud 定价 execution 计算法与本地不一致，迁移要重算成本",[38,6333,6334],{},"复杂 workflow 调试比 Zapier 难（错误堆栈深）",[38,6336,6337],{},"Sustainable Use License 不是传统 OSI 开源，企业法务要看条款",[38,6339,6340],{},"AI Agent 节点 production 稳定性不如简单线性 flow",[38,6342,6343],{},"Webhook 公网暴露要加 IP 白名单 + secret",[38,6345,6346],{},"Worker 模式才能并发，单进程吞吐有限",[20,6348,6350],{"id":6349},"上手docker-5-分钟","上手（Docker 5 分钟）",[3309,6352,6354],{"className":3311,"code":6353,"language":3313,"meta":569,"style":569},"# 持久化目录\nmkdir -p ~\u002Fn8n-data\n\n# 启动\ndocker run -d \\\n  --name n8n \\\n  -p 5678:5678 \\\n  -e N8N_BASIC_AUTH_ACTIVE=true \\\n  -e N8N_BASIC_AUTH_USER=admin \\\n  -e N8N_BASIC_AUTH_PASSWORD=yourpassword \\\n  -v ~\u002Fn8n-data:\u002Fhome\u002Fnode\u002F.n8n \\\n  --restart always \\\n  n8nio\u002Fn8n\n\n# http:\u002F\u002Flocalhost:5678\n",[195,6355,6356,6361,6371,6375,6379,6391,6401,6411,6424,6433,6442,6452,6462,6467,6471],{"__ignoreMap":569},[3317,6357,6358],{"class":3319,"line":3320},[3317,6359,6360],{"class":3369},"# 持久化目录\n",[3317,6362,6363,6366,6368],{"class":3319,"line":573},[3317,6364,6365],{"class":3323},"mkdir",[3317,6367,5781],{"class":3336},[3317,6369,6370],{"class":3327}," ~\u002Fn8n-data\n",[3317,6372,6373],{"class":3319,"line":570},[3317,6374,3515],{"emptyLinePlaceholder":593},[3317,6376,6377],{"class":3319,"line":602},[3317,6378,4463],{"class":3369},[3317,6380,6381,6383,6385,6388],{"class":3319,"line":603},[3317,6382,598],{"class":3323},[3317,6384,5760],{"class":3327},[3317,6386,6387],{"class":3336}," -d",[3317,6389,6390],{"class":3336}," \\\n",[3317,6392,6393,6396,6399],{"class":3319,"line":3512},[3317,6394,6395],{"class":3336},"  --name",[3317,6397,6398],{"class":3327}," n8n",[3317,6400,6390],{"class":3336},[3317,6402,6403,6406,6409],{"class":3319,"line":3518},[3317,6404,6405],{"class":3336},"  -p",[3317,6407,6408],{"class":3327}," 5678:5678",[3317,6410,6390],{"class":3336},[3317,6412,6413,6416,6419,6422],{"class":3319,"line":3524},[3317,6414,6415],{"class":3336},"  -e",[3317,6417,6418],{"class":3327}," N8N_BASIC_AUTH_ACTIVE=",[3317,6420,6421],{"class":3336},"true",[3317,6423,6390],{"class":3336},[3317,6425,6426,6428,6431],{"class":3319,"line":3530},[3317,6427,6415],{"class":3336},[3317,6429,6430],{"class":3327}," N8N_BASIC_AUTH_USER=admin",[3317,6432,6390],{"class":3336},[3317,6434,6435,6437,6440],{"class":3319,"line":3535},[3317,6436,6415],{"class":3336},[3317,6438,6439],{"class":3327}," N8N_BASIC_AUTH_PASSWORD=yourpassword",[3317,6441,6390],{"class":3336},[3317,6443,6444,6447,6450],{"class":3319,"line":3541},[3317,6445,6446],{"class":3336},"  -v",[3317,6448,6449],{"class":3327}," ~\u002Fn8n-data:\u002Fhome\u002Fnode\u002F.n8n",[3317,6451,6390],{"class":3336},[3317,6453,6454,6457,6460],{"class":3319,"line":3546},[3317,6455,6456],{"class":3336},"  --restart",[3317,6458,6459],{"class":3327}," always",[3317,6461,6390],{"class":3336},[3317,6463,6464],{"class":3319,"line":3552},[3317,6465,6466],{"class":3327},"  n8nio\u002Fn8n\n",[3317,6468,6469],{"class":3319,"line":3558},[3317,6470,3515],{"emptyLinePlaceholder":593},[3317,6472,6473],{"class":3319,"line":4485},[3317,6474,6475],{"class":3369},"# http:\u002F\u002Flocalhost:5678\n",[25,6477,6478],{},"连 Ollama：",[3309,6480,6482],{"className":3311,"code":6481,"language":3313,"meta":569,"style":569},"ollama serve\nollama pull llama3.2\n# n8n 中添加 Ollama credential：http:\u002F\u002Fhost.docker.internal:11434\n",[195,6483,6484,6492,6502],{"__ignoreMap":569},[3317,6485,6486,6489],{"class":3319,"line":3320},[3317,6487,6488],{"class":3323},"ollama",[3317,6490,6491],{"class":3327}," serve\n",[3317,6493,6494,6496,6499],{"class":3319,"line":573},[3317,6495,6488],{"class":3323},[3317,6497,6498],{"class":3327}," pull",[3317,6500,6501],{"class":3327}," llama3.2\n",[3317,6503,6504],{"class":3319,"line":570},[3317,6505,6506],{"class":3369},"# n8n 中添加 Ollama credential：http:\u002F\u002Fhost.docker.internal:11434\n",[25,6508,6509],{},"试 workflow：Webhook 触发 → AI Agent 节点（Ollama）→ Slack 通知。复制粘贴一个社区 workflow 30 分钟跑通完整 AI 自动化。",[20,6511,267],{"id":267},[97,6513,6514,6530],{},[100,6515,6516],{},[103,6517,6518,6520,6522,6525,6528],{},[106,6519,276],{},[106,6521,287],{},[106,6523,6524],{},"Zapier",[106,6526,6527],{},"Make",[106,6529,281],{},[115,6531,6532,6545,6557,6574,6586,6600,6612,6626,6642],{},[103,6533,6534,6536,6539,6541,6543],{},[120,6535,2043],{},[120,6537,6538],{},"✅ fair-code",[120,6540,872],{},[120,6542,872],{},[120,6544,4704],{},[103,6546,6547,6549,6551,6553,6555],{},[120,6548,3415],{},[120,6550,5885],{},[120,6552,872],{},[120,6554,872],{},[120,6556,347],{},[103,6558,6559,6562,6565,6568,6571],{},[120,6560,6561],{},"集成数",[120,6563,6564],{},"400+",[120,6566,6567],{},"8000+",[120,6569,6570],{},"2000+",[120,6572,6573],{},"LangChain 原语",[103,6575,6576,6578,6580,6582,6584],{},[120,6577,2352],{},[120,6579,5885],{},[120,6581,5902],{},[120,6583,5902],{},[120,6585,347],{},[103,6587,6588,6591,6594,6596,6598],{},[120,6589,6590],{},"LangChain 节点",[120,6592,6593],{},"✅ 70 个",[120,6595,872],{},[120,6597,872],{},[120,6599,1874],{},[103,6601,6602,6604,6606,6608,6610],{},[120,6603,3923],{},[120,6605,1874],{},[120,6607,872],{},[120,6609,872],{},[120,6611,5902],{},[103,6613,6614,6617,6620,6622,6624],{},[120,6615,6616],{},"Local LLM",[120,6618,6619],{},"✅ Ollama",[120,6621,872],{},[120,6623,872],{},[120,6625,347],{},[103,6627,6628,6631,6634,6637,6640],{},[120,6629,6630],{},"起价",[120,6632,6633],{},"$0 自托管",[120,6635,6636],{},"$29.99\u002F月",[120,6638,6639],{},"$9\u002F月",[120,6641,6633],{},[103,6643,6644,6646,6649,6652,6655],{},[120,6645,398],{},[120,6647,6648],{},"开发者 + 高量级",[120,6650,6651],{},"非技术 + 简单",[120,6653,6654],{},"中等复杂",[120,6656,6657],{},"LangChain 工程",[20,6659,411],{"id":411},[35,6661,6662,6668,6674,6680,6686,6692,6698,6704,6710,6715],{},[38,6663,6664,6667],{},[41,6665,6666],{},"自托管装 Postgres + Redis","：默认 SQLite 高吞吐崩",[38,6669,6670,6673],{},[41,6671,6672],{},"Worker 模式","：高并发要起 worker container 才能并行",[38,6675,6676,6679],{},[41,6677,6678],{},"Webhook 加防护","：公网 Webhook 加 IP 白名单 \u002F secret \u002F nginx",[38,6681,6682,6685],{},[41,6683,6684],{},"数据加密","：n8n encryption key 设强随机值，备份要带 key",[38,6687,6688,6691],{},[41,6689,6690],{},"Cloud vs Self-host 成本","：>2k execution\u002F月 自托管更省",[38,6693,6694,6697],{},[41,6695,6696],{},"集成缺失","：冷门 SaaS 用 HTTP Request + curl 等价",[38,6699,6700,6703],{},[41,6701,6702],{},"AI Agent 稳定性","：生产关键流先用线性节点，agent 留给探索任务",[38,6705,6706,6709],{},[41,6707,6708],{},"license 法务","：Sustainable Use License 给法务看一遍，企业内部用没问题",[38,6711,6712,6714],{},[41,6713,6251],{},"：导入前看作者 + star 数 + 不要直接生产用，要 review",[38,6716,6717,6720],{},[41,6718,6719],{},"monitoring","：生产部署加 prometheus + 错误告警",[20,6722,459],{"id":458},[35,6724,6725,6728,6731,6734,6737,6740,6743,6746],{},[38,6726,6727],{},"✅ 开发者 + 完全控制 + 高量级自动化",[38,6729,6730],{},"✅ 从 Zapier \u002F Make 迁出降本",[38,6732,6733],{},"✅ AI Agent + LangChain + MCP 一体",[38,6735,6736],{},"✅ 合规 \u002F 数据驻留 \u002F 自托管需求",[38,6738,6739],{},"❌ 非技术 + 要 8000+ 现成集成（用 Zapier）",[38,6741,6742],{},"❌ 完全不愿自托管 + 不想付 Cloud",[38,6744,6745],{},"❌ 纯研究 \u002F 学术 agent（用 OpenManus）",[38,6747,6748],{},"❌ 极简 2-step 自动化（Zapier 更快）",[20,6750,526],{"id":526},[35,6752,6753,6758,6762],{},[38,6754,6755],{},[530,6756,6757],{"href":6124},"Langflow 评测",[38,6759,6760],{},[530,6761,6053],{"href":6052},[38,6763,6764],{},[530,6765,6059],{"href":6058},[20,6767,545],{"id":545},[237,6769,6770,6777,6784,6791],{},[38,6771,6772,6773],{},"n8n 官网 ",[530,6774,6775],{"href":6775,"rel":6776},"https:\u002F\u002Fn8n.io",[559],[38,6778,6779,6780],{},"AutomationByExperts — n8n 2026 200k users 5x ARR ",[530,6781,6782],{"href":6782,"rel":6783},"https:\u002F\u002Fautomationbyexperts.com\u002Fblog\u002Fn8n-ai-workflow-automation-guide-2026",[559],[38,6785,6786,6787],{},"Tutorials Technology — n8n + AI on Linux 2026（Docker + Ollama）",[530,6788,6789],{"href":6789,"rel":6790},"https:\u002F\u002Ftutorials.technology\u002Ftutorials\u002Fn8n-ai-workflows-linux-2026.html",[559],[38,6792,6793,6794],{},"Northflank — n8n Self-host Architecture + Pricing 2026 ",[530,6795,6796],{"href":6796,"rel":6797},"https:\u002F\u002Fnorthflank.com\u002Fblog\u002Fhow-to-self-host-n8n-setup-architecture-and-pricing-guide",[559],[3988,6799,6800],{},"html pre.shiki code .sJ8bj, html code.shiki .sJ8bj{--shiki-default:#6A737D;--shiki-dark:#6A737D}html pre.shiki code .sScJk, html code.shiki .sScJk{--shiki-default:#6F42C1;--shiki-dark:#B392F0}html pre.shiki code .sj4cs, html code.shiki .sj4cs{--shiki-default:#005CC5;--shiki-dark:#79B8FF}html pre.shiki code .sZZnC, html code.shiki .sZZnC{--shiki-default:#032F62;--shiki-dark:#9ECBFF}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"title":569,"searchDepth":570,"depth":570,"links":6802},[6803,6804,6805,6806,6807,6808,6809,6810,6811,6812],{"id":22,"depth":573,"text":23},{"id":33,"depth":573,"text":33},{"id":95,"depth":573,"text":95},{"id":6285,"depth":573,"text":6286},{"id":6349,"depth":573,"text":6350},{"id":267,"depth":573,"text":267},{"id":411,"depth":573,"text":411},{"id":458,"depth":573,"text":459},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},"\u002Fimg\u002Ftools\u002Fn8n.webp","n8n 2026 真实评测：开源自托管自动化平台和 AI Agent 工作流工具，支持 LangChain 节点、MCP、Ollama、OpenAI、Webhook、400+ 集成和 execution-based 定价。本文对比 Zapier、Make、Dify，整理自托管成本、适合场景和避坑建议。",[6816,6819,6822,6825],{"q":6817,"a":6818},"n8n 和 Zapier \u002F Make 怎么选？","Zapier 8000+ 集成 + 最易上手 + 非技术团队最爱，但 $29.99\u002F月才 750 tasks + 每步独立计费 = 量大成本爆炸。Make 视觉画布 + 并行分支 + 2000+ 集成 + 智能打包步骤（10k ops $29\u002F月），中等复杂最佳性价比。n8n 开源 + 自托管 + execution-based（步骤数无关）+ 70 LangChain 节点 + MCP 原生，开发者 + 高量级 + 完全控制首选。模式：从 Zapier \u002F Make 起步 → 撞墙 → 迁 n8n。",{"q":6820,"a":6821},"AI Agent 节点是什么？","n8n 2026 加的特殊节点：运行 reasoning loop，从连接的工具节点中自主挑选并调用，直到有答案。和传统线性节点的『按顺序执行预定动作』不同，AI Agent 引入了 LLM 决策。可挂接近 70 LangChain 节点 \u002F MCP server \u002F 任意 HTTP API。让 n8n 从『纯自动化』升级为『真正的 AI agent 平台』。",{"q":6823,"a":6824},"Sustainable Use License 是什么协议？","n8n 用的 fair-code 协议（非传统 OSI 开源）。允许内部使用 + 自托管 + 修改源码，但限制把 n8n 作为 SaaS 转售（与 n8n 商业版直接竞争）。对自用 \u002F 内部工具 \u002F 普通自托管 100% 免费。要做 n8n competitor \u002F 商业 SaaS 才需要谈授权。",{"q":6826,"a":6827},"自托管成本和门槛？","最小：$5-10\u002F月 VPS（DigitalOcean \u002F Hetzner \u002F Linode）+ Docker 一行起。Postgres 持久化 + Redis 队列（高吞吐）+ Worker 节点（水平扩展）。10 分钟内能跑通最小版本。中文社区 \u002F B 站 \u002F 知乎 有大量中文教程。比 Langflow \u002F Dify 上手快。",[590,6122],{},[1111,6126,598,2463],[6832,6836,6840,6844],{"plan":6833,"price":125,"features":6834,"notes":6835},"Community (Self-host)","全部功能 + 不限执行 + 不限 workflow + Sustainable Use License","VPS $5-10\u002F月",{"plan":133,"price":6837,"features":6838,"notes":6839},"~€20\u002F月","2,500 executions + 5 workflow + 基础集成","试水 \u002F 小团队",{"plan":144,"price":6841,"features":6842,"notes":6843},"~€50\u002F月","更高 executions + 团队协作 + 高级特性","中型团队",{"plan":155,"price":158,"features":6845,"notes":6846},"SSO + audit + LDAP + 私有部署 + SLA","大客户","Self-host 免费 \u002F Cloud Starter ~€20·月 (2.5k executions) \u002F Pro \u002F Business \u002F Enterprise 阶梯",[6849],"onboarding\u002Fn8n-ollama-automation",{"power":603,"ux":602,"price":603,"cn_support":570,"stability":603},{"title":287,"description":6814},"n8n 评测 2026：开源自托管自动化平台，AI Agent 工作流首选",[6854,6856,6858,6860],{"name":6855,"url":6775,"accessed":2520},"n8n 官网",{"name":6857,"url":6782,"accessed":2520},"AutomationByExperts — n8n 2026 200k users 5x ARR",{"name":6859,"url":6789,"accessed":2520},"Tutorials Technology — n8n + AI on Linux 2026",{"name":6861,"url":6796,"accessed":2520},"Northflank — n8n Self-host Pricing 2026","tools\u002Fagent\u002Fplatform\u002Fn8n","自托管自动化平台：200k+ 用户 + 70 LangChain 节点 + MCP 原生 + Ollama 集成",[614,2517,6865,617,4058,1111,287],"automation","Zapier \u002F Make 的开源替代——20 步工作流和 2 步成本一样（自托管）。AI Agent + LangChain 节点让它在 2026 成为开发者首选自动化平台。要 8000+ 现成集成 + 极简上手用 Zapier；要中等复杂 + 不自托管用 Make。","Y5uedkjOvG7WLZTjkH2_rw8c5QipFprpPc7QrzBLivw",{"id":6869,"title":542,"alternatives":6870,"api_compatible":15,"body":6871,"category":584,"chinese_friendly":602,"cover":7351,"description":7352,"domestic":587,"extension":588,"faq":15,"free":587,"github":7336,"languages":7353,"lastVerified":591,"meta":7354,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":7355,"pillar":595,"platforms":7356,"priceTable":15,"pricing":599,"published":600,"relatedPlaybooks":15,"relatedReviews":15,"score":7357,"self_host":587,"seo":7358,"seoTitle":7359,"slug":7360,"sources":7361,"stem":7364,"suitable":15,"tagline":7365,"tags":7366,"updated":591,"verdict":7368,"website":7330,"__hash__":7369},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fragflow.md",[13,625,1104],{"type":17,"value":6872,"toc":7338},[6873,6875,6878,6881,6883,6936,6938,6979,6983,6985,6991,7011,7015,7038,7040,7069,7071,7197,7199,7247,7249,7281,7283,7289,7295,7301,7307,7309,7318,7320,7324],[20,6874,23],{"id":22},[25,6876,6877],{},"RAGFlow 是 InfiniFlow（中国团队）出品的开源 RAG 引擎（Apache 2.0），核心卖点是深度文档解析 + 高召回率切片 + 引用溯源。支持 PDF \u002F Word \u002F Excel \u002F PPT \u002F 图片 \u002F 扫描件，内置 OCR + 版面分析 + 表格识别，切片质量远超通用 RAG 方案。Docker 自托管 + Cloud 云端，中文支持好（文档 \u002F UI \u002F 社区）。",[25,6879,6880],{},"适合：需要精准文档问答的企业知识库、复杂文档（表格 \u002F 图文 \u002F 扫描件）场景、中文 RAG 需求、对召回率要求高的业务。不适合：需要复杂 Agent 编排（用 Dify）、资源有限的小服务器、需要精美 UI 的 C 端产品。",[20,6882,33],{"id":33},[35,6884,6885,6891,6897,6903,6909,6915,6920,6925,6931],{},[38,6886,6887,6890],{},[41,6888,6889],{},"深度文档解析","：PDF \u002F Word \u002F Excel \u002F PPT \u002F 图片 \u002F 扫描件，版面分析 + 表格识别",[38,6892,6893,6896],{},[41,6894,6895],{},"OCR 引擎","：内置 PaddleOCR \u002F DeepDOC，支持中英文扫描件识别",[38,6898,6899,6902],{},[41,6900,6901],{},"智能切片","：基于版面分析的语义切片，保留段落 \u002F 表格 \u002F 标题结构",[38,6904,6905,6908],{},[41,6906,6907],{},"高召回率","：混合检索（全文 + 向量）+ 重排序（Rerank），召回精度高",[38,6910,6911,6914],{},[41,6912,6913],{},"引用溯源","：回答标注来源文档 + 页码 + 原文片段，可验证",[38,6916,6917,6919],{},[41,6918,79],{},"：OpenAI \u002F Claude \u002F Ollama \u002F 通义千问 \u002F 智谱 \u002F 月之暗面",[38,6921,6922,6924],{},[41,6923,73],{},"：Elasticsearch \u002F Infinity（自研）\u002F Chroma",[38,6926,6927,6930],{},[41,6928,6929],{},"知识库管理","：多知识库 + 文档分类 + 解析状态监控",[38,6932,6933,6935],{},[41,6934,691],{},"：完整 REST API + SDK，可集成到外部系统",[20,6937,95],{"id":95},[97,6939,6940,6950],{},[100,6941,6942],{},[103,6943,6944,6946,6948],{},[106,6945,108],{},[106,6947,95],{},[106,6949,113],{},[115,6951,6952,6960,6970],{},[103,6953,6954,6956,6958],{},[120,6955,122],{},[120,6957,125],{},[120,6959,128],{},[103,6961,6962,6964,6967],{},[120,6963,722],{},[120,6965,6966],{},"按量付费",[120,6968,6969],{},"托管服务，免运维",[103,6971,6972,6974,6976],{},[120,6973,155],{},[120,6975,158],{},[120,6977,6978],{},"私有部署 + 技术支持 + 定制",[163,6980,6981],{},[25,6982,167],{},[20,6984,171],{"id":170},[163,6986,6987],{},[25,6988,176,6989],{},[41,6990,179],{},[35,6992,6993,6996,6999,7002,7005,7008],{},[38,6994,6995],{},"文档解析质量在开源 RAG 中最强——复杂表格、多栏排版、图文混排都能正确识别",[38,6997,6998],{},"扫描件 OCR 效果好，中文印刷体识别准确率高",[38,7000,7001],{},"引用溯源到页码 + 原文片段，回答可信度高",[38,7003,7004],{},"混合检索 + Rerank 召回精度明显优于纯向量检索",[38,7006,7007],{},"中国团队出品，中文文档和社区支持好，Issue 响应快",[38,7009,7010],{},"支持通义千问 \u002F 智谱 \u002F 月之暗面等国产模型，国内场景适配好",[25,7012,7013],{},[41,7014,209],{},[35,7016,7017,7020,7023,7026,7029,7032,7035],{},[38,7018,7019],{},"资源消耗大——Elasticsearch + Redis + MinIO + RAGFlow 本身，至少 16GB 内存",[38,7021,7022],{},"部署较重，Docker Compose 起来 5+ 容器，配置复杂",[38,7024,7025],{},"大文件解析慢——100 页 PDF 解析 + 切片可能 5-10 分钟",[38,7027,7028],{},"UI 仍有粗糙处，文档管理界面交互不够流畅",[38,7030,7031],{},"Agent 能力弱——RAG 问答是强项，复杂工具调用 \u002F 多步推理不如 Dify",[38,7033,7034],{},"解析失败的重试机制不完善，偶尔卡在 parsing 状态",[38,7036,7037],{},"版本迭代快，升级需注意数据迁移",[20,7039,235],{"id":235},[237,7041,7042,7045,7051,7057,7063,7066],{},[38,7043,7044],{},"系统准备：确保 16GB+ 内存 + Docker + Docker Compose",[38,7046,7047,7048],{},"克隆仓库：",[195,7049,7050],{},"git clone https:\u002F\u002Fgithub.com\u002Finfiniflow\u002Fragflow.git",[38,7052,7053,7054],{},"启动服务：",[195,7055,7056],{},"cd ragflow\u002Fdocker && docker compose up -d",[38,7058,248,7059,7062],{},[195,7060,7061],{},"http:\u002F\u002Flocalhost:80","，注册管理员账号",[38,7064,7065],{},"配置模型：Settings → Model Providers 添加 LLM + Embedding + Rerank",[38,7067,7068],{},"创建知识库 → 上传文档 → 等待解析完成 → 开始问答",[20,7070,267],{"id":267},[97,7072,7073,7087],{},[100,7074,7075],{},[103,7076,7077,7079,7081,7083,7085],{},[106,7078,276],{},[106,7080,542],{},[106,7082,284],{},[106,7084,839],{},[106,7086,623],{},[115,7088,7089,7104,7117,7131,7144,7157,7169,7182],{},[103,7090,7091,7094,7097,7099,7101],{},[120,7092,7093],{},"文档解析",[120,7095,7096],{},"✅ 最强",[120,7098,336],{},[120,7100,897],{},[120,7102,7103],{},"弱",[103,7105,7106,7109,7111,7113,7115],{},[120,7107,7108],{},"表格识别",[120,7110,347],{},[120,7112,872],{},[120,7114,347],{},[120,7116,872],{},[103,7118,7119,7122,7124,7126,7129],{},[120,7120,7121],{},"OCR",[120,7123,2818],{},[120,7125,872],{},[120,7127,7128],{},"需配置",[120,7130,7128],{},[103,7132,7133,7136,7138,7140,7142],{},[120,7134,7135],{},"召回精度",[120,7137,882],{},[120,7139,882],{},[120,7141,882],{},[120,7143,336],{},[103,7145,7146,7148,7151,7153,7155],{},[120,7147,6913],{},[120,7149,7150],{},"✅ 页码+片段",[120,7152,347],{},[120,7154,347],{},[120,7156,347],{},[103,7158,7159,7161,7163,7165,7167],{},[120,7160,679],{},[120,7162,7103],{},[120,7164,897],{},[120,7166,336],{},[120,7168,894],{},[103,7170,7171,7174,7176,7178,7180],{},[120,7172,7173],{},"资源消耗",[120,7175,882],{},[120,7177,336],{},[120,7179,336],{},[120,7181,329],{},[103,7183,7184,7187,7190,7192,7194],{},[120,7185,7186],{},"中文支持",[120,7188,7189],{},"✅ 优秀",[120,7191,1402],{},[120,7193,1402],{},[120,7195,7196],{},"一般",[20,7198,411],{"id":411},[35,7200,7201,7207,7217,7223,7229,7235,7241],{},[38,7202,7203,7206],{},[41,7204,7205],{},"资源一定要够","：低于 16GB 内存别部署，ES + Redis + MinIO 都吃内存",[38,7208,7209,7212,7213,7216],{},[41,7210,7211],{},"Elasticsearch 配置","：默认 JVM 堆偏小，大知识库调 ",[195,7214,7215],{},"ES_JAVA_OPTS"," 到 4-8GB",[38,7218,7219,7222],{},[41,7220,7221],{},"大文件拆分上传","：超过 100 页的 PDF 拆成小文件，解析更稳定",[38,7224,7225,7228],{},[41,7226,7227],{},"解析失败检查格式","：加密 PDF \u002F 损坏文件会卡住，上传前检查",[38,7230,7231,7234],{},[41,7232,7233],{},"Rerank 模型别省","：召回精度提升的关键，用 bge-reranker 或 Cohere Rerank",[38,7236,7237,7240],{},[41,7238,7239],{},"不要当 Agent 平台用","：RAG 问答是核心，复杂工具调用上 Dify",[38,7242,7243,7246],{},[41,7244,7245],{},"定期备份","：ES 数据 + MinIO 文件，升级前完整快照",[20,7248,459],{"id":458},[35,7250,7251,7254,7257,7260,7263,7266,7269,7272,7275,7278],{},[38,7252,7253],{},"✅ 需要精准文档问答的企业知识库",[38,7255,7256],{},"✅ 复杂文档（表格 \u002F 图文 \u002F 扫描件）RAG 场景",[38,7258,7259],{},"✅ 中文 RAG 需求（国产模型 + 中文 OCR）",[38,7261,7262],{},"✅ 对召回率和引用溯源要求高的业务",[38,7264,7265],{},"✅ 有运维能力的团队私有化部署",[38,7267,7268],{},"❌ 需要复杂 Agent 编排（用 Dify）",[38,7270,7271],{},"❌ 资源有限的小服务器（至少 16GB 内存）",[38,7273,7274],{},"❌ 需要精美 C 端 UI 的产品",[38,7276,7277],{},"❌ 无运维能力的团队（用 Cloud 版或 FastGPT）",[38,7279,7280],{},"❌ 纯英文简单文档场景（AnythingLLM 更轻量）",[20,7282,495],{"id":494},[25,7284,7285,7288],{},[41,7286,7287],{},"Q: RAGFlow 和 Dify 怎么选？","\nA: RAGFlow 专注 RAG——文档解析 + 检索精度 + 引用溯源是核心强项，适合文档密集型知识库。Dify 是完整 AI 应用平台——工作流 + Agent + RAG + API 管理，功能更全。纯文档问答选 RAGFlow，构建 AI 应用选 Dify，两者也可配合使用。",[25,7290,7291,7294],{},[41,7292,7293],{},"Q: 部署需要什么配置？","\nA: 最低 16GB 内存 + 4 核 CPU + 50GB 磁盘。生产环境建议 32GB 内存 + 8 核 + SSD。Elasticsearch 是内存大户，知识库文档量大时 ES JVM 堆需 8GB+。如果资源有限，考虑用 Infinity（RAGFlow 自研向量库）替代 ES。",[25,7296,7297,7300],{},[41,7298,7299],{},"Q: 支持中文 OCR 吗？","\nA: 支持。内置 PaddleOCR + DeepDOC 引擎，中文印刷体识别准确率高。手写体效果一般，复杂背景的扫描件建议预处理（去噪 \u002F 矫正）后再上传。OCR 默认开启，可在解析模板中配置。",[25,7302,7303,7306],{},[41,7304,7305],{},"Q: 和 FastGPT 比 RAG 精度如何？","\nA: 两者 RAG 精度都属第一梯队。RAGFlow 的优势在文档解析——复杂表格、多栏版面、图文混排的识别更准确，切片质量更高。FastGPT 的优势在工作流编排和知识库管理 UI 更成熟。文档解析要求高选 RAGFlow，流程管理要求高选 FastGPT。",[20,7308,526],{"id":526},[25,7310,7311,534,7314,534,7316],{},[530,7312,623],{"href":7313},"\u002Fagent\u002Fplatform\u002Fanythingllm.html",[530,7315,10],{"href":1054},[530,7317,1557],{"href":1556},[20,7319,545],{"id":545},[163,7321,7322],{},[25,7323,550],{},[35,7325,7326,7332],{},[38,7327,7328],{},[530,7329,560],{"href":7330,"rel":7331},"https:\u002F\u002Fragflow.io",[559],[38,7333,7334],{},[530,7335,567],{"href":7336,"rel":7337},"https:\u002F\u002Fgithub.com\u002Finfiniflow\u002Fragflow",[559],{"title":569,"searchDepth":570,"depth":570,"links":7339},[7340,7341,7342,7343,7344,7345,7346,7347,7348,7349,7350],{"id":22,"depth":573,"text":23},{"id":33,"depth":573,"text":33},{"id":95,"depth":573,"text":95},{"id":170,"depth":573,"text":171},{"id":235,"depth":573,"text":235},{"id":267,"depth":573,"text":267},{"id":411,"depth":573,"text":411},{"id":458,"depth":573,"text":459},{"id":494,"depth":573,"text":495},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},"\u002Fimg\u002Ftools\u002Fragflow.webp","RAGFlow 真实评测：InfiniFlow 出品的开源 RAG 引擎（Apache 2.0 协议），深度文档解析（PDF\u002FWord\u002FExcel\u002F图片）+ 高召回率切片 + 引用溯源。支持 Docker 自托管，适合需要精准文档问答和知识库检索的企业场景。",[590,2449],{},"\u002Ftools\u002Fagent\u002Fplatform\u002Fragflow",[597,598],{"power":602,"ux":570,"price":603,"cn_support":602,"stability":570},{"title":542,"description":7352},"RAGFlow - 开源 RAG 引擎评测与部署 | AIHO","agent\u002Fplatform\u002Fragflow",[7362,7363],{"title":560,"url":7330},{"title":567,"url":7336},"tools\u002Fagent\u002Fplatform\u002Fragflow","开源 RAG 引擎，深度文档解析 + 高召回率",[613,614,1112,7367,1111],"document-parsing","需要精准文档解析和高召回率 RAG 的企业场景首选，深度文档解析（复杂表格\u002F版面\u002FOCR）+ 引用溯源能力在开源 RAG 引擎中最强，中国团队出品中文支持好，但部署资源要求高、Agent 能力弱、UI 仍需打磨。","bLpjRG4MMrBFYyeFSz4XzLMwVql0SyGD3sB3Tr5g38g",{"id":7371,"title":7372,"alternatives":7373,"api_compatible":15,"body":7374,"category":584,"chinese_friendly":603,"cover":7856,"description":7857,"domestic":587,"extension":588,"faq":7858,"free":587,"github":15,"languages":7871,"lastVerified":15,"meta":7872,"models":15,"navigation":593,"notSuitable":15,"opensource":587,"path":7873,"pillar":595,"platforms":7874,"priceTable":7878,"pricing":7890,"published":6141,"relatedPlaybooks":7891,"relatedReviews":15,"score":7893,"self_host":587,"seo":7894,"seoTitle":7895,"slug":1618,"sources":7896,"stem":7905,"suitable":15,"tagline":7906,"tags":7907,"updated":2520,"verdict":7913,"website":7821,"__hash__":7914},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fyuanqi.md","腾讯元器",[12,14,5546],{"type":17,"value":7375,"toc":7844},[7376,7378,7381,7384,7386,7460,7462,7481,7486,7490,7494,7520,7524,7550,7554,7577,7579,7711,7713,7768,7770,7796,7798,7812,7814],[20,7377,23],{"id":22},[25,7379,7380],{},"腾讯元器（yuanqi.tencent.com）是腾讯官方零代码智能体平台，2024 推出、2026-01 完成系统升级。差异点：『人人可用、零代码、零基础』+ 公众号扫码一键授权（历史群发文章 → 智能体知识库）+ 深度整合 LLM + RAG + 工作流 + 多 Agent 协作 + 一键发布到微信客服 \u002F 公众号 \u002F 服务号 \u002F 企业微信 \u002F 应用宝 + 打通微信支付 MCP 闭环 + 智能体广场 \u002F 模板 \u002F 插件广场 + 混元 + DeepSeek 双底座。",[25,7382,7383],{},"适合：微信公众号博主 \u002F 国内自媒体 + 想做粉丝 7×24 问答；政企客服 \u002F 政务 \u002F 民生场景；电商品牌 + 微信生态 + 微信支付闭环；非技术运营人员 + 零代码上手。不适合：海外用户 \u002F 跨平台分发（用 Coze 海外版 \u002F Dify）；需要自托管 \u002F 数据零云（用 Dify）；纯通用 reasoning \u002F 学术 agent；要深度 API + 工程化（用 Langflow \u002F LangChain）。",[20,7385,33],{"id":33},[35,7387,7388,7394,7400,7406,7412,7418,7424,7430,7436,7442,7448,7454],{},[38,7389,7390,7393],{},[41,7391,7392],{},"公众号一键授权","：扫码绑定 → 历史文章变知识库",[38,7395,7396,7399],{},[41,7397,7398],{},"零代码创建","：自然语言描述 → 智能体 + 工作流",[38,7401,7402,7405],{},[41,7403,7404],{},"混元 + DeepSeek 双底座","：选模型 + 备案合规",[38,7407,7408,7411],{},[41,7409,7410],{},"知识库 RAG","：文章 \u002F PDF \u002F 网页 \u002F 自定义文本",[38,7413,7414,7417],{},[41,7415,7416],{},"工作流引擎","：可视化编排多步执行",[38,7419,7420,7423],{},[41,7421,7422],{},"多 Agent 协作","：子 agent 调子 agent 完成复杂任务",[38,7425,7426,7429],{},[41,7427,7428],{},"插件广场","：搜索 \u002F 天气 \u002F 翻译 \u002F 图片生成等",[38,7431,7432,7435],{},[41,7433,7434],{},"智能体广场 \u002F 模板","：clone 业内成熟智能体",[38,7437,7438,7441],{},[41,7439,7440],{},"全微信生态发布","：客服 \u002F 公众号 \u002F 服务号 \u002F 企业微信 \u002F 应用宝",[38,7443,7444,7447],{},[41,7445,7446],{},"微信支付 MCP","：智能体可直接发起支付闭环",[38,7449,7450,7453],{},[41,7451,7452],{},"企业版能力","：私有部署 + SSO + 定制",[38,7455,7456,7459],{},[41,7457,7458],{},"小助手 \u002F 客服助手 \u002F IP 分身 \u002F 角色陪聊","等内置场景",[20,7461,95],{"id":95},[35,7463,7464,7470,7475],{},[38,7465,7466,7469],{},[41,7467,7468],{},"免费创建","：¥0；创建智能体 + 公众号绑定 + 知识库 + 工作流",[38,7471,7472,7474],{},[41,7473,6966],{},"：混元 \u002F DeepSeek token + 知识库检索 + 工作流执行 + 插件调用按次数",[38,7476,7477,7480],{},[41,7478,7479],{},"企业级方案","：联系商务；私有部署 + SSO + 定制 + SLA",[163,7482,7483],{},[25,7484,7485],{},"实际成本：个人公众号博主一月几十块 token 钱足够跑数千次 7×24 问答；企业客服需求要走商务报价。",[20,7487,7489],{"id":7488},"实测公众号自媒体-政企客服","实测（公众号自媒体 + 政企客服）",[25,7491,7492],{},[41,7493,179],{},[35,7495,7496,7499,7502,7505,7508,7511,7514,7517],{},[38,7497,7498],{},"公众号扫码授权 + 历史文章入库自动化程度极高，5 分钟跑通",[38,7500,7501],{},"混元 + DeepSeek 中文效果好 + 国内备案合规",[38,7503,7504],{},"微信生态全链路：粉丝问 → 智能体答 → 推荐产品 → 微信支付 闭环",[38,7506,7507],{},"智能体广场可看到童爸育儿（2000+ 篇文章库）\u002F 飞哥律界 等真实案例",[38,7509,7510],{},"政务 \u002F 央企（钦州政务 \u002F 中国石化『小石头』）有正式落地",[38,7512,7513],{},"多 Agent 协作架构让复杂业务场景可拆分",[38,7515,7516],{},"零代码运营人员可独立完成，不依赖技术",[38,7518,7519],{},"智能体模板让冷启动快 10x",[25,7521,7522],{},[41,7523,209],{},[35,7525,7526,7529,7532,7535,7538,7541,7544,7547],{},[38,7527,7528],{},"新旧版双底层，旧版用户迁移有摩擦",[38,7530,7531],{},"海外用户访问受限，跨海外业务用不上",[38,7533,7534],{},"工作流编排能力不如 Dify \u002F Coze 强（少高级控制流）",[38,7536,7537],{},"多 Agent 编排 UI 学习曲线",[38,7539,7540],{},"知识库切片粒度自定义有限",[38,7542,7543],{},"国产模型 reasoning 能力比 GPT-5 \u002F Claude 仍有差距",[38,7545,7546],{},"自托管 \u002F 数据私有化不可行（除非企业版私有部署）",[38,7548,7549],{},"海外大模型接入限制（混元 \u002F DeepSeek 为主）",[20,7551,7553],{"id":7552},"上手公众号博主-5-分钟","上手（公众号博主 5 分钟）",[237,7555,7556,7559,7562,7565,7568,7571,7574],{},[38,7557,7558],{},"yuanqi.tencent.com → 微信扫码登录",[38,7560,7561],{},"新建智能体 → 名字 + 头像 + 描述",[38,7563,7564],{},"知识库 → 绑定公众号 → 扫码授权 → 等历史文章入库（几分钟）",[38,7566,7567],{},"角色设定 → 例：『你是 X 公众号官方助手，基于知识库回答问题』",[38,7569,7570],{},"测试对话 → 验证答案命中文章",[38,7572,7573],{},"发布 → 选择渠道（公众号菜单 \u002F 客服）",[38,7575,7576],{},"粉丝在公众号点击菜单 → 进入 AI 对话",[20,7578,267],{"id":267},[97,7580,7581,7596],{},[100,7582,7583],{},[103,7584,7585,7587,7589,7591,7593],{},[106,7586,276],{},[106,7588,2197],{},[106,7590,1616],{},[106,7592,284],{},[106,7594,7595],{},"文心智能体",[115,7597,7598,7610,7624,7640,7654,7667,7684,7697],{},[103,7599,7600,7602,7604,7606,7608],{},[120,7601,7392],{},[120,7603,5885],{},[120,7605,872],{},[120,7607,872],{},[120,7609,5902],{},[103,7611,7612,7615,7618,7620,7622],{},[120,7613,7614],{},"微信支付闭环",[120,7616,7617],{},"✅ MCP",[120,7619,872],{},[120,7621,872],{},[120,7623,872],{},[103,7625,7626,7629,7632,7635,7637],{},[120,7627,7628],{},"多渠道发布",[120,7630,7631],{},"微信生态",[120,7633,7634],{},"多平台 + 海外",[120,7636,3415],{},[120,7638,7639],{},"百度生态",[103,7641,7642,7645,7647,7650,7652],{},[120,7643,7644],{},"海外可用",[120,7646,872],{},[120,7648,7649],{},"✅ Coze.com",[120,7651,347],{},[120,7653,872],{},[103,7655,7656,7658,7661,7663,7665],{},[120,7657,3415],{},[120,7659,7660],{},"❌（企业版才有）",[120,7662,872],{},[120,7664,4695],{},[120,7666,872],{},[103,7668,7669,7672,7675,7678,7681],{},[120,7670,7671],{},"LLM 底座",[120,7673,7674],{},"混元 + DeepSeek",[120,7676,7677],{},"多家",[120,7679,7680],{},"多家 + 自托管",[120,7682,7683],{},"文心一言",[103,7685,7686,7688,7691,7693,7695],{},[120,7687,4741],{},[120,7689,7690],{},"✅ 基础",[120,7692,5885],{},[120,7694,5885],{},[120,7696,347],{},[103,7698,7699,7701,7704,7706,7709],{},[120,7700,398],{},[120,7702,7703],{},"公众号 + 国内",[120,7705,7634],{},[120,7707,7708],{},"自托管 + 企业",[120,7710,7639],{},[20,7712,411],{"id":411},[35,7714,7715,7721,7727,7733,7739,7744,7750,7756,7762],{},[38,7716,7717,7720],{},[41,7718,7719],{},"新版优先","：旧版功能停止演进，新建议直接新版",[38,7722,7723,7726],{},[41,7724,7725],{},"公众号文章质量","：知识库 = 文章质量，垃圾文章入库 = 垃圾回答",[38,7728,7729,7732],{},[41,7730,7731],{},"测试对话","：发布前测 20 个常见问题，验证答案准确",[38,7734,7735,7738],{},[41,7736,7737],{},"兜底回复","：知识库命中失败要有兜底引导（『请联系人工』）",[38,7740,7741,7743],{},[41,7742,7614],{},"：用户体验要清晰，支付链路要可中断",[38,7745,7746,7749],{},[41,7747,7748],{},"token 预算","：开通调用前看模型计价 + 设上限",[38,7751,7752,7755],{},[41,7753,7754],{},"海外用户","：访问可能不稳，业务覆盖海外要走 Coze",[38,7757,7758,7761],{},[41,7759,7760],{},"政务 \u002F 央企合规","：用企业版私有部署 + 备案",[38,7763,7764,7767],{},[41,7765,7766],{},"多 Agent 不要过度","：复杂业务先单 agent 跑通，再拆分",[20,7769,459],{"id":458},[35,7771,7772,7775,7778,7781,7784,7787,7790,7793],{},[38,7773,7774],{},"✅ 微信公众号博主 \u002F 国内自媒体",[38,7776,7777],{},"✅ 政企客服 \u002F 政务 \u002F 民生场景",[38,7779,7780],{},"✅ 电商品牌 + 微信支付闭环",[38,7782,7783],{},"✅ 零代码运营 + 非技术团队",[38,7785,7786],{},"❌ 海外用户 \u002F 跨平台分发",[38,7788,7789],{},"❌ 自托管 \u002F 数据零云",[38,7791,7792],{},"❌ 纯通用 reasoning \u002F 学术 agent",[38,7794,7795],{},"❌ 深度工程化 + API-first",[20,7797,526],{"id":526},[35,7799,7800,7804,7808],{},[38,7801,7802],{},[530,7803,6757],{"href":6124},[38,7805,7806],{},[530,7807,6047],{"href":6046},[38,7809,7810],{},[530,7811,6053],{"href":6052},[20,7813,545],{"id":545},[237,7815,7816,7823,7830,7837],{},[38,7817,7818,7819],{},"腾讯元器官网 ",[530,7820,7821],{"href":7821,"rel":7822},"https:\u002F\u002Fyuanqi.tencent.com",[559],[38,7824,7825,7826],{},"腾讯元器帮助中心 - 平台介绍 + 升级说明 ",[530,7827,7828],{"href":7828,"rel":7829},"https:\u002F\u002Fyuanqi.tencent.com\u002Fguide\u002Fyuanqi-introduction",[559],[38,7831,7832,7833],{},"智能体广场（真实落地案例：童爸育儿 \u002F 飞哥律界 \u002F 钦州政务 \u002F 中国石化小石头）",[530,7834,7835],{"href":7835,"rel":7836},"https:\u002F\u002Fyuanqi.tencent.com\u002Fagent-center\u002Findex",[559],[38,7838,7839,7840],{},"AGI 空间站 — 腾讯混元 + 元器生态 ",[530,7841,7842],{"href":7842,"rel":7843},"https:\u002F\u002Fdocs.feishu.cn\u002Fv\u002Fwiki\u002FJc1TwOaAni879pksMgmcqgUNnof\u002Fa3",[559],{"title":569,"searchDepth":570,"depth":570,"links":7845},[7846,7847,7848,7849,7850,7851,7852,7853,7854,7855],{"id":22,"depth":573,"text":23},{"id":33,"depth":573,"text":33},{"id":95,"depth":573,"text":95},{"id":7488,"depth":573,"text":7489},{"id":7552,"depth":573,"text":7553},{"id":267,"depth":573,"text":267},{"id":411,"depth":573,"text":411},{"id":458,"depth":573,"text":459},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},"\u002Fimg\u002Ftools\u002Fyuanqi.webp","腾讯元器（yuanqi.tencent.com）真实评测：腾讯官方零代码智能体平台，2024 推出、2026-01 完成系统升级（新旧双版本并存）。差异点：『人人可用、零代码、零基础』面向普通用户 + 公众号扫码一键授权（历史群发文章 → 智能体知识库）+ 深度整合 LLM + RAG + 工作流引擎 + 多 Agent 协作架构 + 一键发布到微信客服 \u002F 公众号 \u002F 服务号 \u002F 企业微信 \u002F 应用宝 + 打通微信支付 MCP 闭环 + 智能体广场 \u002F 模板 \u002F 插件广场 + 混元 + DeepSeek 双底座。",[7859,7862,7865,7868],{"q":7860,"a":7861},"和扣子 Coze \u002F Dify \u002F 文心智能体怎么选？","元器的核心独门：『微信公众号一键授权 + 文章变知识库 + 发布到微信全生态 + 微信支付 MCP』。Coze（字节）强『多 LLM + Bot 商店 + 海外版』。Dify 强『LLMOps 全平台 + 可自托管』。文心智能体强『百度生态 + 搜索 + 文心一言』。微信公众号博主 \u002F 政企客户 → 元器；多平台分发 \u002F 跨海外 → Coze；自托管 \u002F 企业 → Dify；百度生态 → 文心。",{"q":7863,"a":7864},"新旧版关系？","2025-12 元器完成系统升级，新版底层逻辑重写（多 Agent 架构 + 微信支付 MCP 等新能力）。旧版创建的智能体 \u002F 知识库 \u002F 工作流暂时无法直接同步到新版（两套底层），需走迁移路径。新建议直接用新版。官方文档：『从旧版元器迁移智能体到新版元器』。",{"q":7866,"a":7867},"公众号绑定后如何运作？","公众号管理员扫码授权 → 元器拉取历史群发文章 → 自动切片 + embedding 入知识库 → 公众号菜单栏出现『智能问答』入口 → 粉丝点击进入智能体对话 → 命中知识库的问题用文章原文 + LLM 综合回答 → 命中失败走兜底回复。这套路径让自媒体几乎零成本拥有 7×24 客服 \u002F 问答机器人。",{"q":7869,"a":7870},"数据隐私和合规？","元器是腾讯官方服务 + 数据托管在腾讯云 + 合规接入备案算法（混元 \u002F DeepSeek）。对国内合规 \u002F 政务 \u002F 央企场景非常友好。但对隐私敏感 \u002F 数据零云需求 → 走 Dify 自托管 + 国产开源模型路径，不要用元器。",[2449,590],{},"\u002Ftools\u002Fagent\u002Fplatform\u002Fyuanqi",[2463,7875,7876,7877],"wechat-mp","wechat-work","wechat-pay",[7879,7882,7886],{"plan":7468,"price":2467,"features":7880,"notes":7881},"创建智能体 + 公众号绑定 + 知识库 + 工作流 + 智能体广场","模型 token 按量",{"plan":6966,"price":7883,"features":7884,"notes":7885},"按调用","混元 \u002F DeepSeek \u002F 知识库检索 \u002F 工作流执行 \u002F 插件调用","token + API 次数",{"plan":7479,"price":7887,"features":7888,"notes":7889},"联系商务","私有部署 + SSO + audit + 定制开发 + 多 agent 协作 + SLA","政企客户","免费创建 \u002F 模型调用按 token \u002F 部分能力按调用次数 \u002F 企业版联系商务",[7892],"onboarding\u002Fwechat-mp-ai-agent",{"power":602,"ux":603,"price":603,"cn_support":603,"stability":602},{"title":7372,"description":7857},"腾讯元器 AI Agent 平台评测 2026：腾讯混元大模型应用开发",[7897,7899,7901,7903],{"name":7898,"url":7821,"accessed":2520},"腾讯元器官网",{"name":7900,"url":7828,"accessed":2520},"腾讯元器帮助中心 - 平台介绍",{"name":7902,"url":7835,"accessed":2520},"智能体广场",{"name":7904,"url":7842,"accessed":2520},"AGI 空间站 — 腾讯混元 + 元器生态","tools\u002Fagent\u002Fplatform\u002Fyuanqi","腾讯零代码智能体平台——公众号一键变 AI 分身 + 混元 \u002F DeepSeek 双底座 + 微信生态全链路",[7908,7909,2516,7910,7911,7912],"agent-builder","wechat","hunyuan","deepseek","yuanqi","微信公众号博主 \u002F 国内自媒体 \u002F 政企客服首选——把公众号变 7×24 AI 分身的最低成本路径。要纯通用 agent \u002F 海外用户 \u002F 复杂 reasoning 用 Dify \u002F Coze。","ecU4jqtN9rSFz0nKLjXaXWfKQf4tCqmUcnzZ5BCUJMI",[],[],[7918,8693,9141],{"id":3023,"title":284,"alternatives":7919,"api_compatible":15,"body":7920,"category":584,"chinese_friendly":602,"cover":4013,"description":4014,"domestic":587,"extension":588,"faq":15,"free":587,"github":3304,"languages":8675,"lastVerified":15,"meta":8676,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":4018,"pillar":595,"platforms":8677,"priceTable":8678,"pricing":4037,"published":2486,"relatedPlaybooks":15,"relatedReviews":8683,"score":8684,"self_host":593,"seo":8685,"seoTitle":4041,"slug":13,"sources":8686,"stem":4054,"suitable":15,"tagline":4055,"tags":8692,"updated":2520,"verdict":4059,"website":3958,"__hash__":4060},[2495,625,14,12],{"type":17,"value":7921,"toc":8653},[7922,7924,7937,7941,7943,7945,7947,8001,8003,8005,8007,8017,8026,8028,8030,8036,8038,8042,8050,8052,8054,8056,8058,8063,8109,8113,8115,8120,8160,8162,8176,8178,8180,8226,8230,8232,8308,8310,8330,8332,8452,8462,8492,8494,8541,8543,8545,8559,8561,8581,8583,8621,8623,8647,8651],[20,7923,23],{"id":22},[1625,7925,7927,7931],{"className":7926},[1628,1629,1630],[25,7928,7929,3036],{},[41,7930,1635],{},[25,7932,1654,7933,3042,7935,3046],{},[41,7934,3041],{},[530,7936,1616],{"href":3045},[163,7938,7939],{},[25,7940,3051],{},[20,7942,1672],{"id":1672},[1674,7944,3057],{"id":3056},[25,7946,3060],{},[97,7948,7949,7959],{},[100,7950,7951],{},[103,7952,7953,7955,7957],{},[106,7954,3069],{},[106,7956,1409],{},[106,7958,3074],{},[115,7960,7961,7971,7981,7991],{},[103,7962,7963,7967,7969],{},[120,7964,7965],{},[41,7966,3083],{},[120,7968,3086],{},[120,7970,3089],{},[103,7972,7973,7977,7979],{},[120,7974,7975],{},[41,7976,358],{},[120,7978,3098],{},[120,7980,3101],{},[103,7982,7983,7987,7989],{},[120,7984,7985],{},[41,7986,3108],{},[120,7988,3111],{},[120,7990,3114],{},[103,7992,7993,7997,7999],{},[120,7994,7995],{},[41,7996,3121],{},[120,7998,3124],{},[120,8000,3127],{},[25,8002,3130],{},[1674,8004,3134],{"id":3133},[25,8006,3137],{},[237,8008,8009,8011,8013,8015],{},[38,8010,3142],{},[38,8012,3145],{},[38,8014,3148],{},[38,8016,3151],{},[25,8018,3154,8019,3160,8022,3164,8024,3167],{},[530,8020,3159],{"href":3157,"rel":8021},[559],[41,8023,3163],{},[530,8025,839],{"href":2030},[1674,8027,3171],{"id":3170},[25,8029,3174],{},[25,8031,3177,8032,3180,8034,3185],{},[530,8033,839],{"href":2030},[530,8035,3184],{"href":3183},[1674,8037,3189],{"id":3188},[25,8039,3192,8040,3197],{},[530,8041,3196],{"href":3195},[35,8043,8044,8046,8048],{},[38,8045,3202],{},[38,8047,3205],{},[38,8049,3208],{},[1674,8051,3212],{"id":3211},[25,8053,3215],{},[20,8055,1932],{"id":1932},[1674,8057,3221],{"id":3220},[25,8059,3224,8060,3230],{},[530,8061,3229],{"href":3227,"rel":8062},[559],[97,8064,8065,8075],{},[100,8066,8067],{},[103,8068,8069,8071,8073],{},[106,8070,3239],{},[106,8072,95],{},[106,8074,3244],{},[115,8076,8077,8085,8093,8101],{},[103,8078,8079,8081,8083],{},[120,8080,3251],{},[120,8082,3254],{},[120,8084,3257],{},[103,8086,8087,8089,8091],{},[120,8088,3262],{},[120,8090,3265],{},[120,8092,3268],{},[103,8094,8095,8097,8099],{},[120,8096,3273],{},[120,8098,3276],{},[120,8100,3279],{},[103,8102,8103,8105,8107],{},[120,8104,155],{},[120,8106,158],{},[120,8108,3288],{},[25,8110,3291,8111,3295],{},[41,8112,3294],{},[1674,8114,3299],{"id":3298},[25,8116,8117,3307],{},[530,8118,3306],{"href":3304,"rel":8119},[559],[3309,8121,8122],{"className":3311,"code":3312,"language":3313,"meta":569,"style":569},[195,8123,8124,8132,8138,8146,8156],{"__ignoreMap":569},[3317,8125,8126,8128,8130],{"class":3319,"line":3320},[3317,8127,3324],{"class":3323},[3317,8129,3328],{"class":3327},[3317,8131,3331],{"class":3327},[3317,8133,8134,8136],{"class":3319,"line":573},[3317,8135,3337],{"class":3336},[3317,8137,3340],{"class":3327},[3317,8139,8140,8142,8144],{"class":3319,"line":570},[3317,8141,3345],{"class":3323},[3317,8143,3348],{"class":3327},[3317,8145,3351],{"class":3327},[3317,8147,8148,8150,8152,8154],{"class":3319,"line":602},[3317,8149,598],{"class":3323},[3317,8151,3358],{"class":3327},[3317,8153,3361],{"class":3327},[3317,8155,3364],{"class":3336},[3317,8157,8158],{"class":3319,"line":603},[3317,8159,3370],{"class":3369},[25,8161,3373],{},[35,8163,8164,8168,8172],{},[38,8165,8166,3381],{},[41,8167,3380],{},[38,8169,8170,3387],{},[41,8171,3386],{},[38,8173,8174,3393],{},[41,8175,3392],{},[1674,8177,3397],{"id":3396},[25,8179,3400],{},[97,8181,8182,8192],{},[100,8183,8184],{},[103,8185,8186,8188,8190],{},[106,8187,3409],{},[106,8189,3412],{},[106,8191,3415],{},[115,8193,8194,8202,8210,8218],{},[103,8195,8196,8198,8200],{},[120,8197,3422],{},[120,8199,3425],{},[120,8201,125],{},[103,8203,8204,8206,8208],{},[120,8205,3432],{},[120,8207,125],{},[120,8209,3437],{},[103,8211,8212,8214,8216],{},[120,8213,3442],{},[120,8215,3445],{},[120,8217,3448],{},[103,8219,8220,8222,8224],{},[120,8221,3453],{},[120,8223,3456],{},[120,8225,3459],{},[25,8227,8228,3465],{},[41,8229,3464],{},[20,8231,1968],{"id":1967},[3309,8233,8234],{"className":3311,"code":3470,"language":3313,"meta":569,"style":569},[195,8235,8236,8240,8248,8254,8262,8272,8276,8280,8284,8288,8292,8296,8300,8304],{"__ignoreMap":569},[3317,8237,8238],{"class":3319,"line":3320},[3317,8239,3477],{"class":3369},[3317,8241,8242,8244,8246],{"class":3319,"line":573},[3317,8243,3324],{"class":3323},[3317,8245,3328],{"class":3327},[3317,8247,3331],{"class":3327},[3317,8249,8250,8252],{"class":3319,"line":570},[3317,8251,3337],{"class":3336},[3317,8253,3340],{"class":3327},[3317,8255,8256,8258,8260],{"class":3319,"line":602},[3317,8257,3345],{"class":3323},[3317,8259,3348],{"class":3327},[3317,8261,3351],{"class":3327},[3317,8263,8264,8266,8268,8270],{"class":3319,"line":603},[3317,8265,598],{"class":3323},[3317,8267,3358],{"class":3327},[3317,8269,3361],{"class":3327},[3317,8271,3364],{"class":3336},[3317,8273,8274],{"class":3319,"line":3512},[3317,8275,3515],{"emptyLinePlaceholder":593},[3317,8277,8278],{"class":3319,"line":3518},[3317,8279,3521],{"class":3369},[3317,8281,8282],{"class":3319,"line":3524},[3317,8283,3527],{"class":3369},[3317,8285,8286],{"class":3319,"line":3530},[3317,8287,3515],{"emptyLinePlaceholder":593},[3317,8289,8290],{"class":3319,"line":3535},[3317,8291,3538],{"class":3369},[3317,8293,8294],{"class":3319,"line":3541},[3317,8295,3515],{"emptyLinePlaceholder":593},[3317,8297,8298],{"class":3319,"line":3546},[3317,8299,3549],{"class":3369},[3317,8301,8302],{"class":3319,"line":3552},[3317,8303,3555],{"class":3369},[3317,8305,8306],{"class":3319,"line":3558},[3317,8307,3561],{"class":3369},[20,8309,3564],{"id":3564},[237,8311,8312,8316,8320,8324],{},[38,8313,8314,3572],{},[41,8315,3571],{},[38,8317,8318,3578],{},[41,8319,3577],{},[38,8321,8322,3584],{},[41,8323,3583],{},[38,8325,8326,3590,8328,3593],{},[41,8327,3589],{},[530,8329,1616],{"href":3045},[20,8331,2010],{"id":2010},[97,8333,8334,8354],{},[100,8335,8336],{},[103,8337,8338,8340,8342,8346,8350],{},[106,8339,276],{},[106,8341,284],{},[106,8343,8344],{},[530,8345,1616],{"href":3045},[106,8347,8348],{},[530,8349,839],{"href":2030},[106,8351,8352],{},[530,8353,287],{"href":2345},[115,8355,8356,8368,8380,8392,8404,8416,8428,8440],{},[103,8357,8358,8360,8362,8364,8366],{},[120,8359,2043],{},[120,8361,347],{},[120,8363,872],{},[120,8365,347],{},[120,8367,3632],{},[103,8369,8370,8372,8374,8376,8378],{},[120,8371,2056],{},[120,8373,347],{},[120,8375,872],{},[120,8377,347],{},[120,8379,347],{},[103,8381,8382,8384,8386,8388,8390],{},[120,8383,1334],{},[120,8385,2095],{},[120,8387,3653],{},[120,8389,2095],{},[120,8391,2089],{},[103,8393,8394,8396,8398,8400,8402],{},[120,8395,2086],{},[120,8397,2092],{},[120,8399,2089],{},[120,8401,2095],{},[120,8403,2092],{},[103,8405,8406,8408,8410,8412,8414],{},[120,8407,877],{},[120,8409,2089],{},[120,8411,2095],{},[120,8413,2092],{},[120,8415,3682],{},[103,8417,8418,8420,8422,8424,8426],{},[120,8419,3687],{},[120,8421,2092],{},[120,8423,2089],{},[120,8425,3694],{},[120,8427,2089],{},[103,8429,8430,8432,8434,8436,8438],{},[120,8431,3701],{},[120,8433,2089],{},[120,8435,2092],{},[120,8437,2089],{},[120,8439,2095],{},[103,8441,8442,8444,8446,8448,8450],{},[120,8443,3714],{},[120,8445,872],{},[120,8447,3719],{},[120,8449,872],{},[120,8451,872],{},[25,8453,8454,3728,8456,2162,8459,3735],{},[41,8455,2155],{},[530,8457,2161],{"href":2159,"rel":8458},[559],[530,8460,2166],{"href":1661,"rel":8461},[559],[35,8463,8464,8468,8474,8480,8486],{},[38,8465,8466,3743],{},[41,8467,3742],{},[38,8469,8470,2180,8472],{},[41,8471,3748],{},[530,8473,1616],{"href":3045},[38,8475,8476,2180,8478],{},[41,8477,2187],{},[530,8479,839],{"href":2030},[38,8481,8482,2180,8484],{},[41,8483,3761],{},[530,8485,287],{"href":2345},[38,8487,8488,2180,8490],{},[41,8489,3768],{},[530,8491,281],{"href":2326},[20,8493,2200],{"id":2200},[35,8495,8496,8500,8510,8517,8525,8529,8533,8537],{},[38,8497,8498,3780],{},[41,8499,3779],{},[38,8501,8502,2892,8506,3792,8508,3796],{},[41,8503,8504,3788],{},[195,8505,3787],{},[195,8507,3791],{},[195,8509,3795],{},[38,8511,8512,2892,8514,3807],{},[41,8513,3801],{},[530,8515,3806],{"href":3804,"rel":8516},[559],[38,8518,8519,3813,8521,3816,8523,3820],{},[41,8520,3812],{},[195,8522,3787],{},[195,8524,3819],{},[38,8526,8527,3826],{},[41,8528,3825],{},[38,8530,8531,3832],{},[41,8532,3831],{},[38,8534,8535,3838],{},[41,8536,3837],{},[38,8538,8539,3844],{},[41,8540,3843],{},[20,8542,459],{"id":458},[25,8544,2278],{},[35,8546,8547,8549,8551,8553,8555,8557],{},[38,8548,3853],{},[38,8550,3856],{},[38,8552,3859],{},[38,8554,3862],{},[38,8556,3865],{},[38,8558,3868],{},[25,8560,2298],{},[35,8562,8563,8567,8571,8573,8577],{},[38,8564,3875,8565,3878],{},[530,8566,1616],{"href":3045},[38,8568,3881,8569,3884],{},[530,8570,839],{"href":2030},[38,8572,3887],{},[38,8574,3890,8575,3893],{},[530,8576,1616],{"href":3045},[38,8578,3896,8579,3899],{},[530,8580,287],{"href":2345},[20,8582,526],{"id":526},[35,8584,8585,8595,8605,8615],{},[38,8586,2334,8587,2258,8589,2258,8591,2258,8593],{},[530,8588,1616],{"href":3045},[530,8590,839],{"href":2030},[530,8592,287],{"href":2345},[530,8594,281],{"href":2326},[38,8596,3916,8597,2258,8599,2258,8601,2258,8603],{},[530,8598,2352],{"href":2351},[530,8600,341],{"href":2355},[530,8602,3923],{"href":3195},[530,8604,2359],{"href":2358},[38,8606,3928,8607,2258,8609,2258,8611,2258,8613],{},[530,8608,3932],{"href":3931},[530,8610,3936],{"href":3935},[530,8612,2373],{"href":2372},[530,8614,2381],{"href":2380},[38,8616,2384,8617,2258,8619],{},[530,8618,3946],{"href":3945},[530,8620,2392],{"href":2391},[20,8622,545],{"id":545},[35,8624,8625,8630,8635,8640,8645],{},[38,8626,3955,8627],{},[530,8628,3958],{"href":3958,"rel":8629},[559],[38,8631,3962,8632],{},[530,8633,3804],{"href":3804,"rel":8634},[559],[38,8636,3968,8637],{},[530,8638,3304],{"href":3304,"rel":8639},[559],[38,8641,3974,8642],{},[530,8643,3977],{"href":3977,"rel":8644},[559],[38,8646,3981],{},[25,8648,3984,8649,2429],{},[530,8650,2428],{"href":2427},[3988,8652,3990],{},{"title":569,"searchDepth":570,"depth":570,"links":8654},[8655,8656,8663,8668,8669,8670,8671,8672,8673,8674],{"id":22,"depth":573,"text":23},{"id":1672,"depth":573,"text":1672,"children":8657},[8658,8659,8660,8661,8662],{"id":3056,"depth":570,"text":3057},{"id":3133,"depth":570,"text":3134},{"id":3170,"depth":570,"text":3171},{"id":3188,"depth":570,"text":3189},{"id":3211,"depth":570,"text":3212},{"id":1932,"depth":573,"text":1932,"children":8664},[8665,8666,8667],{"id":3220,"depth":570,"text":3221},{"id":3298,"depth":570,"text":3299},{"id":3396,"depth":570,"text":3397},{"id":1967,"depth":573,"text":1968},{"id":3564,"depth":573,"text":3564},{"id":2010,"depth":573,"text":2010},{"id":2200,"depth":573,"text":2200},{"id":458,"depth":573,"text":459},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},[2449,590,4016],{},[1098,1099,597,598],[8679,8680,8681,8682],{"plan":4022,"price":125,"features":4023,"notes":4024},{"plan":4026,"price":125,"features":4027,"notes":4028},{"plan":4030,"price":3265,"features":4031,"notes":4032},{"plan":4034,"price":4035,"features":4036,"notes":158},[2488,2489,2490,2491],{"power":603,"ux":602,"price":603,"cn_support":602,"stability":602},{"title":284,"description":4014},[8687,8688,8689,8690,8691],{"title":4044,"url":3804},{"title":4046,"url":3304},{"title":4048,"url":3977},{"title":4050,"url":2159},{"title":4052,"url":4053},[613,614,1111,1112,2517,4057,4058],{"id":5544,"title":281,"alternatives":8694,"api_compatible":15,"body":8695,"category":584,"chinese_friendly":570,"cover":6106,"description":6107,"domestic":587,"extension":588,"faq":9119,"free":587,"github":15,"languages":9124,"lastVerified":15,"meta":9125,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":6124,"pillar":595,"platforms":9126,"priceTable":9127,"pricing":6140,"published":6141,"relatedPlaybooks":9132,"relatedReviews":15,"score":9133,"self_host":593,"seo":9134,"seoTitle":6146,"slug":12,"sources":9135,"stem":6157,"suitable":15,"tagline":6158,"tags":9140,"updated":2520,"verdict":6162,"website":6150,"__hash__":6163},[14,5546,5547],{"type":17,"value":8696,"toc":9107},[8697,8699,8701,8703,8705,8755,8757,8775,8779,8781,8785,8803,8807,8825,8827,8871,8873,8889,8891,9005,9007,9045,9047,9065,9067,9081,9083,9105],[20,8698,23],{"id":22},[25,8700,5554],{},[25,8702,5557],{},[20,8704,33],{"id":33},[35,8706,8707,8711,8715,8719,8723,8727,8731,8735,8739,8743,8747,8751],{},[38,8708,8709,5567],{},[41,8710,5566],{},[38,8712,8713,5573],{},[41,8714,5572],{},[38,8716,8717,5579],{},[41,8718,5578],{},[38,8720,8721,5585],{},[41,8722,5584],{},[38,8724,8725,5591],{},[41,8726,5590],{},[38,8728,8729,5597],{},[41,8730,5596],{},[38,8732,8733,5603],{},[41,8734,5602],{},[38,8736,8737,5609],{},[41,8738,5608],{},[38,8740,8741,5615],{},[41,8742,5614],{},[38,8744,8745,5621],{},[41,8746,5620],{},[38,8748,8749,5627],{},[41,8750,5626],{},[38,8752,8753,5633],{},[41,8754,5632],{},[20,8756,95],{"id":95},[35,8758,8759,8763,8767,8771],{},[38,8760,8761,5643],{},[41,8762,5642],{},[38,8764,8765,5649],{},[41,8766,5648],{},[38,8768,8769,5655],{},[41,8770,5654],{},[38,8772,8773,5660],{},[41,8774,155],{},[163,8776,8777],{},[25,8778,5665],{},[20,8780,5669],{"id":5668},[25,8782,8783],{},[41,8784,179],{},[35,8786,8787,8789,8791,8793,8795,8797,8799,8801],{},[38,8788,5678],{},[38,8790,5681],{},[38,8792,5684],{},[38,8794,5687],{},[38,8796,5690],{},[38,8798,5693],{},[38,8800,5696],{},[38,8802,5699],{},[25,8804,8805],{},[41,8806,209],{},[35,8808,8809,8811,8813,8815,8817,8819,8821,8823],{},[38,8810,5708],{},[38,8812,5711],{},[38,8814,5714],{},[38,8816,5717],{},[38,8818,5720],{},[38,8820,5723],{},[38,8822,5726],{},[38,8824,5729],{},[20,8826,235],{"id":235},[3309,8828,8829],{"className":3311,"code":5734,"language":3313,"meta":569,"style":569},[195,8830,8831,8835,8843,8851,8855,8859],{"__ignoreMap":569},[3317,8832,8833],{"class":3319,"line":3320},[3317,8834,5741],{"class":3369},[3317,8836,8837,8839,8841],{"class":3319,"line":573},[3317,8838,5746],{"class":3323},[3317,8840,5749],{"class":3327},[3317,8842,5752],{"class":3327},[3317,8844,8845,8847,8849],{"class":3319,"line":570},[3317,8846,5757],{"class":3323},[3317,8848,5760],{"class":3327},[3317,8850,5763],{"class":3369},[3317,8852,8853],{"class":3319,"line":602},[3317,8854,3515],{"emptyLinePlaceholder":593},[3317,8856,8857],{"class":3319,"line":603},[3317,8858,5772],{"class":3369},[3317,8860,8861,8863,8865,8867,8869],{"class":3319,"line":3512},[3317,8862,598],{"class":3323},[3317,8864,5760],{"class":3327},[3317,8866,5781],{"class":3336},[3317,8868,5784],{"class":3327},[3317,8870,5787],{"class":3327},[25,8872,5790],{},[237,8874,8875,8877,8879,8881,8883,8885,8887],{},[38,8876,5795],{},[38,8878,5798],{},[38,8880,5801],{},[38,8882,5804],{},[38,8884,5807],{},[38,8886,5810],{},[38,8888,5813],{},[20,8890,267],{"id":267},[97,8892,8893,8907],{},[100,8894,8895],{},[103,8896,8897,8899,8901,8903,8905],{},[106,8898,276],{},[106,8900,281],{},[106,8902,284],{},[106,8904,287],{},[106,8906,10],{},[115,8908,8909,8921,8933,8945,8957,8969,8981,8993],{},[103,8910,8911,8913,8915,8917,8919],{},[120,8912,5838],{},[120,8914,5841],{},[120,8916,5844],{},[120,8918,5847],{},[120,8920,5850],{},[103,8922,8923,8925,8927,8929,8931],{},[120,8924,2043],{},[120,8926,4704],{},[120,8928,5859],{},[120,8930,5862],{},[120,8932,4704],{},[103,8934,8935,8937,8939,8941,8943],{},[120,8936,3415],{},[120,8938,5871],{},[120,8940,5874],{},[120,8942,5874],{},[120,8944,347],{},[103,8946,8947,8949,8951,8953,8955],{},[120,8948,2787],{},[120,8950,5885],{},[120,8952,347],{},[120,8954,347],{},[120,8956,347],{},[103,8958,8959,8961,8963,8965,8967],{},[120,8960,5896],{},[120,8962,5899],{},[120,8964,5902],{},[120,8966,5905],{},[120,8968,5905],{},[103,8970,8971,8973,8975,8977,8979],{},[120,8972,5912],{},[120,8974,347],{},[120,8976,347],{},[120,8978,5902],{},[120,8980,347],{},[103,8982,8983,8985,8987,8989,8991],{},[120,8984,5925],{},[120,8986,5928],{},[120,8988,5931],{},[120,8990,5934],{},[120,8992,5937],{},[103,8994,8995,8997,8999,9001,9003],{},[120,8996,398],{},[120,8998,5944],{},[120,9000,5947],{},[120,9002,322],{},[120,9004,5952],{},[20,9006,411],{"id":411},[35,9008,9009,9013,9017,9021,9025,9029,9033,9037,9041],{},[38,9010,9011,5962],{},[41,9012,5961],{},[38,9014,9015,5968],{},[41,9016,5967],{},[38,9018,9019,5974],{},[41,9020,5973],{},[38,9022,9023,5980],{},[41,9024,5979],{},[38,9026,9027,5986],{},[41,9028,5985],{},[38,9030,9031,5992],{},[41,9032,5991],{},[38,9034,9035,5998],{},[41,9036,5997],{},[38,9038,9039,6003],{},[41,9040,3701],{},[38,9042,9043,6009],{},[41,9044,6008],{},[20,9046,459],{"id":458},[35,9048,9049,9051,9053,9055,9057,9059,9061,9063],{},[38,9050,6016],{},[38,9052,6019],{},[38,9054,6022],{},[38,9056,6025],{},[38,9058,6028],{},[38,9060,6031],{},[38,9062,6034],{},[38,9064,6037],{},[20,9066,526],{"id":526},[35,9068,9069,9073,9077],{},[38,9070,9071],{},[530,9072,6047],{"href":6046},[38,9074,9075],{},[530,9076,6053],{"href":6052},[38,9078,9079],{},[530,9080,6059],{"href":6058},[20,9082,545],{"id":545},[237,9084,9085,9090,9095,9100],{},[38,9086,6066,9087],{},[530,9088,6069],{"href":6069,"rel":9089},[559],[38,9091,6073,9092],{},[530,9093,6076],{"href":6076,"rel":9094},[559],[38,9096,6080,9097],{},[530,9098,6083],{"href":6083,"rel":9099},[559],[38,9101,6087,9102],{},[530,9103,6090],{"href":6090,"rel":9104},[559],[3988,9106,5048],{},{"title":569,"searchDepth":570,"depth":570,"links":9108},[9109,9110,9111,9112,9113,9114,9115,9116,9117,9118],{"id":22,"depth":573,"text":23},{"id":33,"depth":573,"text":33},{"id":95,"depth":573,"text":95},{"id":5668,"depth":573,"text":5669},{"id":235,"depth":573,"text":235},{"id":267,"depth":573,"text":267},{"id":411,"depth":573,"text":411},{"id":458,"depth":573,"text":459},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},[9120,9121,9122,9123],{"q":6110,"a":6111},{"q":6113,"a":6114},{"q":6116,"a":6117},{"q":6119,"a":6120},[590,6122],{},[1111,6126,598,2463],[9128,9129,9130,9131],{"plan":5642,"price":125,"features":6129,"notes":6130},{"plan":5648,"price":125,"features":6132,"notes":5090},{"plan":5654,"price":6134,"features":6135,"notes":6136},{"plan":155,"price":158,"features":6138,"notes":6139},[6143],{"power":602,"ux":603,"price":603,"cn_support":570,"stability":602},{"title":281,"description":6107},[9136,9137,9138,9139],{"name":6149,"url":6150,"accessed":2520},{"name":6152,"url":6076,"accessed":2520},{"name":6154,"url":6083,"accessed":2520},{"name":6156,"url":6090,"accessed":2520},[614,6160,617,1112,6161,5757],{"id":6165,"title":287,"alternatives":9142,"api_compatible":15,"body":9143,"category":584,"chinese_friendly":570,"cover":6813,"description":6814,"domestic":587,"extension":588,"faq":9651,"free":587,"github":15,"languages":9656,"lastVerified":15,"meta":9657,"models":15,"navigation":593,"notSuitable":15,"opensource":593,"path":6046,"pillar":595,"platforms":9658,"priceTable":9659,"pricing":6847,"published":6141,"relatedPlaybooks":9664,"relatedReviews":15,"score":9665,"self_host":593,"seo":9666,"seoTitle":6852,"slug":14,"sources":9667,"stem":6862,"suitable":15,"tagline":6863,"tags":9672,"updated":2520,"verdict":6866,"website":6775,"__hash__":6867},[12,5546,5547],{"type":17,"value":9144,"toc":9639},[9145,9147,9149,9151,9153,9203,9205,9223,9227,9229,9233,9251,9255,9273,9275,9379,9381,9403,9405,9407,9533,9535,9577,9579,9597,9599,9613,9615,9637],[20,9146,23],{"id":22},[25,9148,6173],{},[25,9150,6176],{},[20,9152,33],{"id":33},[35,9154,9155,9159,9163,9167,9171,9175,9179,9183,9187,9191,9195,9199],{},[38,9156,9157,6186],{},[41,9158,6185],{},[38,9160,9161,6192],{},[41,9162,6191],{},[38,9164,9165,6198],{},[41,9166,6197],{},[38,9168,9169,6204],{},[41,9170,6203],{},[38,9172,9173,6210],{},[41,9174,6209],{},[38,9176,9177,6216],{},[41,9178,6215],{},[38,9180,9181,6222],{},[41,9182,6221],{},[38,9184,9185,6228],{},[41,9186,6227],{},[38,9188,9189,6234],{},[41,9190,6233],{},[38,9192,9193,6240],{},[41,9194,6239],{},[38,9196,9197,6246],{},[41,9198,6245],{},[38,9200,9201,6252],{},[41,9202,6251],{},[20,9204,95],{"id":95},[35,9206,9207,9211,9215,9219],{},[38,9208,9209,6262],{},[41,9210,6261],{},[38,9212,9213,6267],{},[41,9214,133],{},[38,9216,9217,6272],{},[41,9218,144],{},[38,9220,9221,6277],{},[41,9222,155],{},[163,9224,9225],{},[25,9226,6282],{},[20,9228,6286],{"id":6285},[25,9230,9231],{},[41,9232,179],{},[35,9234,9235,9237,9239,9241,9243,9245,9247,9249],{},[38,9236,6295],{},[38,9238,6298],{},[38,9240,6301],{},[38,9242,6304],{},[38,9244,6307],{},[38,9246,6310],{},[38,9248,6313],{},[38,9250,6316],{},[25,9252,9253],{},[41,9254,209],{},[35,9256,9257,9259,9261,9263,9265,9267,9269,9271],{},[38,9258,6325],{},[38,9260,6328],{},[38,9262,6331],{},[38,9264,6334],{},[38,9266,6337],{},[38,9268,6340],{},[38,9270,6343],{},[38,9272,6346],{},[20,9274,6350],{"id":6349},[3309,9276,9277],{"className":3311,"code":6353,"language":3313,"meta":569,"style":569},[195,9278,9279,9283,9291,9295,9299,9309,9317,9325,9335,9343,9351,9359,9367,9371,9375],{"__ignoreMap":569},[3317,9280,9281],{"class":3319,"line":3320},[3317,9282,6360],{"class":3369},[3317,9284,9285,9287,9289],{"class":3319,"line":573},[3317,9286,6365],{"class":3323},[3317,9288,5781],{"class":3336},[3317,9290,6370],{"class":3327},[3317,9292,9293],{"class":3319,"line":570},[3317,9294,3515],{"emptyLinePlaceholder":593},[3317,9296,9297],{"class":3319,"line":602},[3317,9298,4463],{"class":3369},[3317,9300,9301,9303,9305,9307],{"class":3319,"line":603},[3317,9302,598],{"class":3323},[3317,9304,5760],{"class":3327},[3317,9306,6387],{"class":3336},[3317,9308,6390],{"class":3336},[3317,9310,9311,9313,9315],{"class":3319,"line":3512},[3317,9312,6395],{"class":3336},[3317,9314,6398],{"class":3327},[3317,9316,6390],{"class":3336},[3317,9318,9319,9321,9323],{"class":3319,"line":3518},[3317,9320,6405],{"class":3336},[3317,9322,6408],{"class":3327},[3317,9324,6390],{"class":3336},[3317,9326,9327,9329,9331,9333],{"class":3319,"line":3524},[3317,9328,6415],{"class":3336},[3317,9330,6418],{"class":3327},[3317,9332,6421],{"class":3336},[3317,9334,6390],{"class":3336},[3317,9336,9337,9339,9341],{"class":3319,"line":3530},[3317,9338,6415],{"class":3336},[3317,9340,6430],{"class":3327},[3317,9342,6390],{"class":3336},[3317,9344,9345,9347,9349],{"class":3319,"line":3535},[3317,9346,6415],{"class":3336},[3317,9348,6439],{"class":3327},[3317,9350,6390],{"class":3336},[3317,9352,9353,9355,9357],{"class":3319,"line":3541},[3317,9354,6446],{"class":3336},[3317,9356,6449],{"class":3327},[3317,9358,6390],{"class":3336},[3317,9360,9361,9363,9365],{"class":3319,"line":3546},[3317,9362,6456],{"class":3336},[3317,9364,6459],{"class":3327},[3317,9366,6390],{"class":3336},[3317,9368,9369],{"class":3319,"line":3552},[3317,9370,6466],{"class":3327},[3317,9372,9373],{"class":3319,"line":3558},[3317,9374,3515],{"emptyLinePlaceholder":593},[3317,9376,9377],{"class":3319,"line":4485},[3317,9378,6475],{"class":3369},[25,9380,6478],{},[3309,9382,9383],{"className":3311,"code":6481,"language":3313,"meta":569,"style":569},[195,9384,9385,9391,9399],{"__ignoreMap":569},[3317,9386,9387,9389],{"class":3319,"line":3320},[3317,9388,6488],{"class":3323},[3317,9390,6491],{"class":3327},[3317,9392,9393,9395,9397],{"class":3319,"line":573},[3317,9394,6488],{"class":3323},[3317,9396,6498],{"class":3327},[3317,9398,6501],{"class":3327},[3317,9400,9401],{"class":3319,"line":570},[3317,9402,6506],{"class":3369},[25,9404,6509],{},[20,9406,267],{"id":267},[97,9408,9409,9423],{},[100,9410,9411],{},[103,9412,9413,9415,9417,9419,9421],{},[106,9414,276],{},[106,9416,287],{},[106,9418,6524],{},[106,9420,6527],{},[106,9422,281],{},[115,9424,9425,9437,9449,9461,9473,9485,9497,9509,9521],{},[103,9426,9427,9429,9431,9433,9435],{},[120,9428,2043],{},[120,9430,6538],{},[120,9432,872],{},[120,9434,872],{},[120,9436,4704],{},[103,9438,9439,9441,9443,9445,9447],{},[120,9440,3415],{},[120,9442,5885],{},[120,9444,872],{},[120,9446,872],{},[120,9448,347],{},[103,9450,9451,9453,9455,9457,9459],{},[120,9452,6561],{},[120,9454,6564],{},[120,9456,6567],{},[120,9458,6570],{},[120,9460,6573],{},[103,9462,9463,9465,9467,9469,9471],{},[120,9464,2352],{},[120,9466,5885],{},[120,9468,5902],{},[120,9470,5902],{},[120,9472,347],{},[103,9474,9475,9477,9479,9481,9483],{},[120,9476,6590],{},[120,9478,6593],{},[120,9480,872],{},[120,9482,872],{},[120,9484,1874],{},[103,9486,9487,9489,9491,9493,9495],{},[120,9488,3923],{},[120,9490,1874],{},[120,9492,872],{},[120,9494,872],{},[120,9496,5902],{},[103,9498,9499,9501,9503,9505,9507],{},[120,9500,6616],{},[120,9502,6619],{},[120,9504,872],{},[120,9506,872],{},[120,9508,347],{},[103,9510,9511,9513,9515,9517,9519],{},[120,9512,6630],{},[120,9514,6633],{},[120,9516,6636],{},[120,9518,6639],{},[120,9520,6633],{},[103,9522,9523,9525,9527,9529,9531],{},[120,9524,398],{},[120,9526,6648],{},[120,9528,6651],{},[120,9530,6654],{},[120,9532,6657],{},[20,9534,411],{"id":411},[35,9536,9537,9541,9545,9549,9553,9557,9561,9565,9569,9573],{},[38,9538,9539,6667],{},[41,9540,6666],{},[38,9542,9543,6673],{},[41,9544,6672],{},[38,9546,9547,6679],{},[41,9548,6678],{},[38,9550,9551,6685],{},[41,9552,6684],{},[38,9554,9555,6691],{},[41,9556,6690],{},[38,9558,9559,6697],{},[41,9560,6696],{},[38,9562,9563,6703],{},[41,9564,6702],{},[38,9566,9567,6709],{},[41,9568,6708],{},[38,9570,9571,6714],{},[41,9572,6251],{},[38,9574,9575,6720],{},[41,9576,6719],{},[20,9578,459],{"id":458},[35,9580,9581,9583,9585,9587,9589,9591,9593,9595],{},[38,9582,6727],{},[38,9584,6730],{},[38,9586,6733],{},[38,9588,6736],{},[38,9590,6739],{},[38,9592,6742],{},[38,9594,6745],{},[38,9596,6748],{},[20,9598,526],{"id":526},[35,9600,9601,9605,9609],{},[38,9602,9603],{},[530,9604,6757],{"href":6124},[38,9606,9607],{},[530,9608,6053],{"href":6052},[38,9610,9611],{},[530,9612,6059],{"href":6058},[20,9614,545],{"id":545},[237,9616,9617,9622,9627,9632],{},[38,9618,6772,9619],{},[530,9620,6775],{"href":6775,"rel":9621},[559],[38,9623,6779,9624],{},[530,9625,6782],{"href":6782,"rel":9626},[559],[38,9628,6786,9629],{},[530,9630,6789],{"href":6789,"rel":9631},[559],[38,9633,6793,9634],{},[530,9635,6796],{"href":6796,"rel":9636},[559],[3988,9638,6800],{},{"title":569,"searchDepth":570,"depth":570,"links":9640},[9641,9642,9643,9644,9645,9646,9647,9648,9649,9650],{"id":22,"depth":573,"text":23},{"id":33,"depth":573,"text":33},{"id":95,"depth":573,"text":95},{"id":6285,"depth":573,"text":6286},{"id":6349,"depth":573,"text":6350},{"id":267,"depth":573,"text":267},{"id":411,"depth":573,"text":411},{"id":458,"depth":573,"text":459},{"id":526,"depth":573,"text":526},{"id":545,"depth":573,"text":545},[9652,9653,9654,9655],{"q":6817,"a":6818},{"q":6820,"a":6821},{"q":6823,"a":6824},{"q":6826,"a":6827},[590,6122],{},[1111,6126,598,2463],[9660,9661,9662,9663],{"plan":6833,"price":125,"features":6834,"notes":6835},{"plan":133,"price":6837,"features":6838,"notes":6839},{"plan":144,"price":6841,"features":6842,"notes":6843},{"plan":155,"price":158,"features":6845,"notes":6846},[6849],{"power":603,"ux":602,"price":603,"cn_support":570,"stability":603},{"title":287,"description":6814},[9668,9669,9670,9671],{"name":6855,"url":6775,"accessed":2520},{"name":6857,"url":6782,"accessed":2520},{"name":6859,"url":6789,"accessed":2520},{"name":6861,"url":6796,"accessed":2520},[614,2517,6865,617,4058,1111,287],1785428435778]