[{"data":1,"prerenderedAt":2857},["ShallowReactive",2],{"header-counts":3,"footer-counts":6,"tool-tools\u002Fcoding\u002Flocal\u002Flobe-chat":9,"tool-stats-coding\u002Flocal\u002Flobe-chat":597,"cat-rank-coding-local":600,"tool-related-coding\u002Flocal\u002Flobe-chat":2856},{"tools":4,"reviews":5},65,7,{"tools":4,"reviews":5,"playbooks":7,"news":8},10,8,{"id":10,"title":11,"alternatives":12,"api_compatible":25,"body":26,"category":524,"chinese_friendly":525,"cover":526,"description":527,"domestic":528,"extension":529,"faq":530,"free":528,"github":25,"languages":543,"meta":546,"models":25,"navigation":547,"notSuitable":25,"opensource":547,"path":548,"pillar":549,"platforms":550,"priceTable":556,"pricing":565,"published":566,"relatedPlaybooks":567,"relatedReviews":25,"score":573,"self_host":547,"seo":575,"slug":576,"sources":577,"stem":585,"suitable":25,"tagline":586,"tags":587,"updated":580,"verdict":594,"website":595,"__hash__":596},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Flobe-chat.md","LobeChat",[13,16,19,22],{"name":14,"url":15},"cherry-studio","\u002Ftools\u002Fcoding\u002Flocal\u002Fcherry-studio",{"name":17,"url":18},"open-webui","\u002Ftools\u002Fcoding\u002Flocal\u002Fopen-webui",{"name":20,"url":21},"ollama","\u002Ftools\u002Fcoding\u002Flocal\u002Follama",{"name":23,"url":24},"lm-studio","\u002Ftools\u002Fcoding\u002Flocal\u002Flm-studio",null,{"type":27,"value":28,"toc":509},"minimark",[29,34,43,46,49,117,120,133,136,140,145,165,170,187,190,217,220,384,387,425,429,455,458,482,485],[30,31,33],"h2",{"id":32},"tldr","TL;DR",[35,36,37,38,42],"p",{},"LobeChat 是 LobeHub 团队的开源 AI 聊天框架，2023 年发布、GitHub 72k+ stars、MIT 协议。",[39,40,41],"strong",{},"Web + 桌面 + Docker 自托管三形态","，把 OpenAI \u002F Claude \u002F Gemini \u002F DeepSeek \u002F Qwen \u002F Kimi \u002F Ollama \u002F LM Studio 等 80+ 模型聚合到一个现代设计的客户端里。内置 RAG 知识库 + 插件市场 + 助手市场 + 多模型对比 + MCP，是当下综合最强的多模型 AI 客户端之一。",[35,44,45],{},"适合：需要 Web 端访问、Docker 自托管、多模型对比、丰富助手市场的用户；中文重度用户；想给团队 \u002F 家庭部署一个共享 AI 工作台。不适合：只用桌面 + 不需要 Web（Cherry Studio 同样优秀且更精细）、强企业 RBAC + 多租户（Open WebUI 多用户更完善）。",[30,47,48],{"id":48},"核心能力",[50,51,52,59,65,71,77,83,89,94,100,106],"ul",{},[53,54,55,58],"li",{},[39,56,57],{},"多模型聚合","：OpenAI \u002F Claude \u002F Gemini \u002F DeepSeek \u002F Qwen \u002F Kimi \u002F 豆包 \u002F Groq \u002F Together \u002F OpenRouter \u002F Ollama \u002F LM Studio",[53,60,61,64],{},[39,62,63],{},"多模型对比","：同 prompt 给多模型并排回答",[53,66,67,70],{},[39,68,69],{},"本地 RAG 知识库","：上传 PDF \u002F Word \u002F 网页 → 向量化 → 检索引用",[53,72,73,76],{},[39,74,75],{},"插件市场","：联网搜索 \u002F 代码执行 \u002F 图像生成 \u002F 翻译等几十款官方插件",[53,78,79,82],{},[39,80,81],{},"助手市场","：几百个预设 AI 角色，一键导入",[53,84,85,88],{},[39,86,87],{},"MCP 协议","：扩展任意工具能力",[53,90,91],{},[39,92,93],{},"代码解释器 \u002F 文件上传 \u002F TTS \u002F 多模态",[53,95,96,99],{},[39,97,98],{},"Web + 桌面 + Docker","：三形态，数据可完全本地",[53,101,102,105],{},[39,103,104],{},"LobeHub Cloud","：官方云托管，免部署",[53,107,108,111,112,116],{},[39,109,110],{},"快捷指令 \u002F 工作流","：自定义 prompt 模板，",[113,114,115],"code",{},"\u002Fpodcast-summary"," 类用法",[30,118,119],{"id":119},"价格",[50,121,122,128],{},[53,123,124,127],{},[39,125,126],{},"自托管 \u002F 桌面","：完全免费、MIT 开源",[53,129,130,132],{},[39,131,104],{},"：订阅制，云端托管 + 团队协作 + 同步",[35,134,135],{},"模型 API 费用按你自己的供应商付费；本地 Ollama \u002F LM Studio 零成本。",[30,137,139],{"id":138},"实测m2-自托管-docker连-openai-deepseek-本地-ollama","实测（M2 + 自托管 Docker，连 OpenAI + DeepSeek + 本地 Ollama）",[35,141,142],{},[39,143,144],{},"亮点：",[50,146,147,150,153,156,159,162],{},[53,148,149],{},"界面颜值是这一类工具里第一档（深色 \u002F 透明 \u002F 现代感）",[53,151,152],{},"多模型并排对比对选型极其有用：写一道复杂题，Claude \u002F GPT \u002F DeepSeek 直接对比答案",[53,154,155],{},"知识库 RAG 上传 50+ PDF 后检索准确，引用片段可视化",[53,157,158],{},"助手市场拿来即用——「Code Reviewer」「Translation Polish」节省 prompt 编写",[53,160,161],{},"Docker 一键部署，团队 5 人共享流畅",[53,163,164],{},"多平台数据同步（Cloud \u002F WebDAV）",[35,166,167],{},[39,168,169],{},"踩坑：",[50,171,172,175,178,181,184],{},[53,173,174],{},"自托管要熟悉 Docker + 反代 + HTTPS",[53,176,177],{},"国内连 OpenAI \u002F Claude 需自带网络方案",[53,179,180],{},"Web 版数据存 LobeHub，隐私敏感场景走桌面 \u002F Docker",[53,182,183],{},"插件市场质量参差，要自筛",[53,185,186],{},"团队多人共享需配 LobeHub Cloud 或自建数据库（Postgres + S3）",[30,188,189],{"id":189},"上手",[191,192,193,196,202,205,208,211,214],"ol",{},[53,194,195],{},"选形态：Web（chat.lobehub.com 注册即用） \u002F 桌面（GitHub Releases 下载） \u002F Docker",[53,197,198,199],{},"Docker：",[113,200,201],{},"docker run -d -p 3210:3210 -e OPENAI_API_KEY=sk-xxx --name lobe-chat lobehub\u002Flobe-chat",[53,203,204],{},"设置 → AI 服务商 → 添加 OpenAI \u002F Claude \u002F DeepSeek \u002F Ollama",[53,206,207],{},"模型选择器测试对话",[53,209,210],{},"知识库：拖文件 → 等向量化 → 对话引用",[53,212,213],{},"助手市场拉「Code Reviewer」「论文翻译润色」试用",[53,215,216],{},"进阶：插件市场启用联网搜索 \u002F 代码执行；MCP 自定义工具",[30,218,219],{"id":219},"对比",[221,222,223,244],"table",{},[224,225,226],"thead",{},[227,228,229,233,235,238,241],"tr",{},[230,231,232],"th",{},"维度",[230,234,11],{},[230,236,237],{},"Cherry Studio",[230,239,240],{},"Open WebUI",[230,242,243],{},"LM Studio",[245,246,247,263,278,292,306,322,336,351,368],"tbody",{},[227,248,249,253,255,258,261],{},[250,251,252],"td",{},"形态",[250,254,98],{},[250,256,257],{},"桌面",[250,259,260],{},"Docker \u002F 桌面",[250,262,257],{},[227,264,265,267,270,273,275],{},[250,266,57],{},[250,268,269],{},"✅ 80+",[250,271,272],{},"✅",[250,274,272],{},[250,276,277],{},"本地为主",[227,279,280,282,285,287,290],{},[250,281,63],{},[250,283,284],{},"✅ 一等",[250,286,272],{},[250,288,289],{},"弱",[250,291,289],{},[227,293,294,297,299,301,304],{},[250,295,296],{},"知识库 RAG",[250,298,272],{},[250,300,272],{},[250,302,303],{},"✅ + oikb",[250,305,289],{},[227,307,308,311,314,317,320],{},[250,309,310],{},"插件 \u002F 助手市场",[250,312,313],{},"✅ 丰富",[250,315,316],{},"300+ 助手",[250,318,319],{},"Tools",[250,321,289],{},[227,323,324,327,329,331,334],{},[250,325,326],{},"MCP",[250,328,272],{},[250,330,272],{},[250,332,333],{},"✅ mcpo",[250,335,289],{},[227,337,338,341,344,347,349],{},[250,339,340],{},"多用户",[250,342,343],{},"配 Cloud \u002F 自建",[250,345,346],{},"无",[250,348,284],{},[250,350,346],{},[227,352,353,356,359,362,365],{},[250,354,355],{},"GitHub Stars",[250,357,358],{},"72k+",[250,360,361],{},"60k+",[250,363,364],{},"126k+",[250,366,367],{},"–",[227,369,370,373,376,379,381],{},[250,371,372],{},"开源协议",[250,374,375],{},"MIT",[250,377,378],{},"AGPL-3.0",[250,380,375],{},[250,382,383],{},"闭源（免费）",[30,385,386],{"id":386},"避坑",[50,388,389,395,401,407,413,419],{},[53,390,391,394],{},[39,392,393],{},"Web 版数据不本地","：隐私敏感选桌面或 Docker 自托管",[53,396,397,400],{},[39,398,399],{},"国内连海外模型走中转","：直连 OpenAI \u002F Claude 不稳，配 OpenRouter \u002F Ofox \u002F 国内中转",[53,402,403,406],{},[39,404,405],{},"Docker 自托管暴露公网","：上反代 + HTTPS + Auth + 备份数据库",[53,408,409,412],{},[39,410,411],{},"嵌入模型中文优化","：默认嵌入对中文一般，配 bge-m3 \u002F 硅基流动 Pro 版",[53,414,415,418],{},[39,416,417],{},"插件市场审一遍","：第三方插件可执行代码，团队部署谨慎启用",[53,420,421,424],{},[39,422,423],{},"同步选 Cloud vs WebDAV","：团队多端走 LobeHub Cloud；个人多设备 WebDAV 即可",[30,426,428],{"id":427},"适合-不适合","适合 \u002F 不适合",[50,430,431,434,437,440,443,446,449,452],{},[53,432,433],{},"✅ Web + 桌面双形态需求",[53,435,436],{},"✅ Docker 自托管 \u002F 团队共享",[53,438,439],{},"✅ 多模型对比 \u002F 选型",[53,441,442],{},"✅ 中文重度用户",[53,444,445],{},"✅ 助手市场 \u002F 插件生态用户",[53,447,448],{},"❌ 强企业 RBAC + 多租户（Open WebUI 更完善）",[53,450,451],{},"❌ 只要桌面 + 数据完全本地（Cherry Studio 同样优秀）",[53,453,454],{},"❌ 完全不会碰 Docker",[30,456,457],{"id":457},"相关阅读",[50,459,460,466,471,476],{},[53,461,462],{},[463,464,465],"a",{"href":15},"Cherry Studio 评测",[53,467,468],{},[463,469,470],{"href":18},"Open WebUI 评测",[53,472,473],{},[463,474,475],{"href":21},"Ollama 评测",[53,477,478],{},[463,479,481],{"href":480},"\u002Fplaybook\u002Fonboarding\u002Frag-pipeline-build","RAG Pipeline 搭建 Playbook",[30,483,484],{"id":484},"来源",[191,486,487,495,502],{},[53,488,489,490],{},"LobeChat GitHub 仓库（72k+ stars，MIT）",[463,491,492],{"href":492,"rel":493},"https:\u002F\u002Fgithub.com\u002Flobehub\u002Flobe-chat",[494],"nofollow",[53,496,497,498],{},"腾讯云开发者社区 — Lobe Chat 本地化 AI 聊天终极桌面客户端（2026-01）",[463,499,500],{"href":500,"rel":501},"https:\u002F\u002Fcloud.tencent.com\u002Fdeveloper\u002Farticle\u002F2622150",[494],[53,503,504,505],{},"Ofox.ai — LobeChat 完全配置指南 2026（2026-04-17）",[463,506,507],{"href":507,"rel":508},"https:\u002F\u002Fofox.ai\u002Fzh\u002Fblog\u002Flobechat-api-configuration-guide-2026",[494],{"title":510,"searchDepth":511,"depth":511,"links":512},"",3,[513,515,516,517,518,519,520,521,522,523],{"id":32,"depth":514,"text":33},2,{"id":48,"depth":514,"text":48},{"id":119,"depth":514,"text":119},{"id":138,"depth":514,"text":139},{"id":189,"depth":514,"text":189},{"id":219,"depth":514,"text":219},{"id":386,"depth":514,"text":386},{"id":427,"depth":514,"text":428},{"id":457,"depth":514,"text":457},{"id":484,"depth":514,"text":484},"local",5,"\u002Fimg\u002Ftools\u002Flobe-chat.webp","LobeChat 真实评测：LobeHub 团队开源 AI 聊天框架，GitHub 72k+ stars、MIT 协议。Web + 桌面（Win\u002FMac\u002FLinux\u002FDocker）双形态，支持 OpenAI \u002F Claude \u002F Gemini \u002F DeepSeek \u002F Qwen \u002F Ollama 等 80+ 模型，内置 RAG 知识库 + 插件市场 + 助手市场 + 多模型对比。",false,"md",[531,534,537,540],{"q":532,"a":533},"Web 版 vs 桌面版 vs Docker 自托管，怎么选？","Web 版（chat.lobehub.com）最快上手但数据存 LobeHub 服务器；桌面版数据本地存、隐私好；Docker 自托管对团队 \u002F 公司部署最优，完全掌控数据。",{"q":535,"a":536},"支持哪些模型？","80+ 模型：OpenAI 全系列、Anthropic Claude、Google Gemini、DeepSeek、Qwen、Kimi、Moonshot、字节豆包、Groq、Together、OpenRouter、Ollama \u002F LM Studio 本地模型，以及任何 OpenAI 兼容 API。",{"q":538,"a":539},"多模型对比怎么用？","同一对话窗口里把消息广播给多个模型并排回答，选型 \u002F 评估特别有用——直接看 Claude 和 GPT 在同一 prompt 下的回答差异。",{"q":541,"a":542},"助手市场是什么？","LobeHub 维护的预设 AI 角色市场（代码审查 \u002F 翻译 \u002F 写作 \u002F 角色扮演等几百个），一键拉到本地用，省去自己写 System Prompt。",[544,545],"zh","en",{},true,"\u002Ftools\u002Fcoding\u002Flocal\u002Flobe-chat","coding",[551,552,553,554,555],"web","windows","macos","linux","docker",[557,561],{"plan":126,"price":558,"features":559,"notes":560},"免费","全功能 \u002F 80+ 模型 \u002F 知识库 \u002F 插件 \u002F 助手市场","MIT 协议",{"plan":104,"price":562,"features":563,"notes":564},"订阅制","云端托管 \u002F 免部署 \u002F 团队协作 \u002F 同步","chat.lobehub.com 注册即用","完全免费（MIT 开源） \u002F LobeHub Cloud 订阅","2026-06-19",[568,570],{"name":569,"url":480},"RAG Pipeline 搭建",{"name":571,"url":572},"Claude Code 上手","\u002Fplaybook\u002Fonboarding\u002Fclaude-code-getting-started",{"power":525,"ux":525,"price":525,"cn_support":525,"stability":574},4,{"title":11,"description":527},"coding\u002Flocal\u002Flobe-chat",[578,581,583],{"name":579,"url":492,"accessed":580},"LobeChat GitHub","2026-06-24",{"name":582,"url":500,"accessed":580},"腾讯云开发者社区 — Lobe Chat 终极桌面客户端",{"name":584,"url":507,"accessed":580},"Ofox.ai — LobeChat 完全配置指南 2026","tools\u002Fcoding\u002Flocal\u002Flobe-chat","现代设计的开源 AI 聊天框架——Web + 桌面双形态、72k+ stars、多模型 + 知识库 + 插件市场",[524,551,588,589,590,591,592,593],"desktop","multi-model","rag","plugin","mcp","open-source","颜值与功能双优的多模型 AI 聊天客户端。要 Web + 桌面双形态、自托管 Docker、多模型对比、丰富助手市场——LobeChat 是综合最强；纯桌面体验 Cherry Studio 同样优秀。","https:\u002F\u002Flobehub.com","qiaN3oNudbNSX66toalN5tVx-_meCcKL6o6rq4RdAdI",{"ok":547,"slug":598,"viewCount":599,"clickCount":599,"avgRating":599,"ratingCount":599},"coding%2Flocal%2Flobe-chat",0,[601,1050,1535,1915,2386],{"id":602,"title":237,"alternatives":603,"api_compatible":25,"body":609,"category":524,"chinese_friendly":525,"cover":999,"description":1000,"domestic":528,"extension":529,"faq":1001,"free":528,"github":25,"languages":1014,"meta":1015,"models":25,"navigation":547,"notSuitable":25,"opensource":547,"path":15,"pillar":549,"platforms":1016,"priceTable":1018,"pricing":1026,"published":566,"relatedPlaybooks":1027,"relatedReviews":25,"score":1032,"self_host":547,"seo":1033,"slug":1034,"sources":1035,"stem":1042,"suitable":25,"tagline":1043,"tags":1044,"updated":580,"verdict":1047,"website":1048,"__hash__":1049},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Fcherry-studio.md",[604,606,607,608],{"name":605,"url":548},"lobe-chat",{"name":23,"url":24},{"name":20,"url":21},{"name":17,"url":18},{"type":27,"value":610,"toc":987},[611,613,616,619,621,662,664,678,684,688,692,709,713,730,732,756,758,881,883,915,917,940,942,962,964],[30,612,33],{"id":32},[35,614,615],{},"Cherry Studio 是一款开源、跨平台（Windows \u002F macOS \u002F Linux \u002F Android）的桌面 AI 客户端，定位『全能 AI 工作台』：把 OpenAI \u002F Anthropic \u002F Google \u002F DeepSeek 等云端模型，以及 Ollama \u002F LM Studio 本地模型，全部聚合到同一个桌面应用里管理。内置 300+ 助手模板、本地 RAG 知识库、Markdown + Mermaid 渲染、MCP 协议支持，所有对话数据本地存储 + WebDAV 备份。AGPL-3.0 开源、GitHub 60k+ stars，企业版可联系商务做私有化部署。",[35,617,618],{},"适合：中文 AI 重度用户、想统一管理多家模型、需要本地知识库 RAG、关注数据本地存储的开发者 \u002F 研究者。不适合：要 Web 端访问 \u002F Docker 自托管 \u002F 团队多人共享 \u002F iOS 端使用。",[30,620,48],{"id":48},[50,622,623,628,633,639,644,650,656],{},[53,624,625,627],{},[39,626,57],{},"：OpenAI \u002F Claude \u002F Gemini \u002F DeepSeek \u002F Qwen \u002F Kimi \u002F Moonshot 等云端 + Ollama \u002F LM Studio 本地",[53,629,630,632],{},[39,631,69],{},"：拖拽 PDF \u002F Word \u002F Excel \u002F PPT \u002F 网址 \u002F sitemap → 自动向量化 → 检索增强问答 + 来源追溯",[53,634,635,638],{},[39,636,637],{},"300+ 助手模板","：编程 \u002F 写作 \u002F 翻译 \u002F 学习 \u002F 角色扮演开箱即用，可自定义 System Prompt",[53,640,641,643],{},[39,642,87],{},"：扩展工具调用 \u002F 联网搜索 \u002F 文件操作",[53,645,646,649],{},[39,647,648],{},"数据本地优先","：对话历史本地存储，WebDAV 同步，不上传第三方",[53,651,652,655],{},[39,653,654],{},"多模态","：图片识别 \u002F PDF 阅读 \u002F Markdown + Mermaid + 代码高亮",[53,657,658,661],{},[39,659,660],{},"AI 绘画 + 翻译","：内置主流 SD \u002F DALL·E \u002F 翻译 API 集成",[30,663,119],{"id":119},[50,665,666,672],{},[53,667,668,671],{},[39,669,670],{},"开源版","：完全免费，AGPL-3.0",[53,673,674,677],{},[39,675,676],{},"Enterprise","：私有化部署 + 团队协作 + 资源管控，联系销售",[679,680,681],"blockquote",{},[35,682,683],{},"模型 API 费用按你自己绑定的供应商计费；本地 Ollama \u002F LM Studio 零成本。",[30,685,687],{"id":686},"实测mac-m2-中型知识库","实测（Mac M2 + 中型知识库）",[35,689,690],{},[39,691,144],{},[50,693,694,697,700,703,706],{},[53,695,696],{},"中文 UI \u002F 文档 \u002F 社区都顶级，零门槛上手",[53,698,699],{},"本地 RAG 拖入 30+ PDF 后向量化 \u003C 2 分钟（用 bge-m3）",[53,701,702],{},"多模型并排回答：让 Claude \u002F GPT \u002F DeepSeek 同回一个问题做比较",[53,704,705],{},"MCP 接 Brave Search + 自定义工具流畅",[53,707,708],{},"WebDAV 同步坚果云 \u002F 阿里云盘，桌面 + 移动设备数据互通",[35,710,711],{},[39,712,169],{},[50,714,715,718,721,724,727],{},[53,716,717],{},"没有 Web 端 \u002F Docker 自托管（要这个用 LobeChat）",[53,719,720],{},"iOS 版尚未发布（roadmap 中）",[53,722,723],{},"大型 PDF（>100 MB）向量化偶有失败，要切小",[53,725,726],{},"助手市场质量参差，要自筛",[53,728,729],{},"模型 API 调用全靠你自己付费，新手要先理解 API Key 概念",[30,731,189],{"id":189},[191,733,734,737,740,747,750,753],{},[53,735,736],{},"cherry-ai.com 下载客户端（或 GitHub releases）",[53,738,739],{},"设置 → 模型服务 → 填 OpenAI \u002F Claude \u002F DeepSeek API Key",[53,741,742,743],{},"（可选）本地：装 Ollama → Cherry Studio 自动识别 endpoint ",[463,744,745],{"href":745,"rel":746},"http:\u002F\u002Flocalhost:11434",[494],[53,748,749],{},"新建知识库 → 拖文件 \u002F 加网址 → 等向量化",[53,751,752],{},"新对话 → 选模型 → 勾知识库 → 提问",[53,754,755],{},"进阶：自定义助手（System Prompt）+ MCP 扩展工具",[30,757,219],{"id":219},[221,759,760,774],{},[224,761,762],{},[227,763,764,766,768,770,772],{},[230,765,232],{},[230,767,237],{},[230,769,11],{},[230,771,243],{},[230,773,240],{},[245,775,776,789,802,815,827,842,857,869],{},[227,777,778,780,782,785,787],{},[250,779,252],{},[250,781,257],{},[250,783,784],{},"Web + 桌面",[250,786,257],{},[250,788,260],{},[227,790,791,793,796,798,800],{},[250,792,57],{},[250,794,795],{},"✅ 云 + 本地",[250,797,795],{},[250,799,277],{},[250,801,795],{},[227,803,804,806,809,811,813],{},[250,805,296],{},[250,807,808],{},"✅ 强",[250,810,808],{},[250,812,289],{},[250,814,272],{},[227,816,817,819,821,823,825],{},[250,818,326],{},[250,820,272],{},[250,822,272],{},[250,824,289],{},[250,826,272],{},[227,828,829,832,835,838,840],{},[250,830,831],{},"自托管 \u002F Web",[250,833,834],{},"无 Web",[250,836,837],{},"✅ Docker",[250,839,346],{},[250,841,837],{},[227,843,844,847,850,852,855],{},[250,845,846],{},"中文",[250,848,849],{},"5\u002F5",[250,851,849],{},[250,853,854],{},"4\u002F5",[250,856,854],{},[227,858,859,861,863,865,867],{},[250,860,372],{},[250,862,378],{},[250,864,375],{},[250,866,383],{},[250,868,375],{},[227,870,871,873,875,877,879],{},[250,872,355],{},[250,874,361],{},[250,876,358],{},[250,878,367],{},[250,880,364],{},[30,882,386],{"id":386},[50,884,885,891,897,903,909],{},[53,886,887,890],{},[39,888,889],{},"API Key 别明文外泄","：客户端配置文件以明文存 Key，机器借出前先清；团队共享用企业版 \u002F 自建中转",[53,892,893,896],{},[39,894,895],{},"知识库别一次塞太多","：单库 1000+ 文档检索质量明显下降，按主题切分多个知识库",[53,898,899,902],{},[39,900,901],{},"嵌入模型选择","：免费 bge-m3 够用；专业用付费 Pro\u002FBAAI\u002Fbge-m3 或 OpenAI text-embedding-3",[53,904,905,908],{},[39,906,907],{},"WebDAV 同步先小范围测","：知识库向量数据较大，先备份对话再开同步",[53,910,911,914],{},[39,912,913],{},"MCP 工具来源要可控","：MCP 是给 AI 真实工具能力，第三方插件审一遍代码",[30,916,428],{"id":427},[50,918,919,922,925,928,931,934,937],{},[53,920,921],{},"✅ 中文用户、AI 重度使用 \u002F 多模型管理",[53,923,924],{},"✅ 需要本地 RAG 知识库",[53,926,927],{},"✅ 关注数据隐私 \u002F 本地存储",[53,929,930],{},"✅ 想用 Ollama \u002F LM Studio 本地模型",[53,932,933],{},"❌ 需要 Web 端 \u002F Docker 自托管",[53,935,936],{},"❌ 团队多人共享 \u002F SSO",[53,938,939],{},"❌ iOS 主力用户",[30,941,457],{"id":457},[50,943,944,949,954,958],{},[53,945,946],{},[463,947,948],{"href":548},"LobeChat 评测",[53,950,951],{},[463,952,953],{"href":24},"LM Studio 评测",[53,955,956],{},[463,957,475],{"href":21},[53,959,960],{},[463,961,481],{"href":480},[30,963,484],{"id":484},[191,965,966,973,980],{},[53,967,968,969],{},"Cherry Studio 官网（功能 + 下载）",[463,970,971],{"href":971,"rel":972},"https:\u002F\u002Fwww.cherry-ai.com\u002F",[494],[53,974,975,976],{},"MBLUO Studio — Cherry Studio 评测 2026 ",[463,977,978],{"href":978,"rel":979},"https:\u002F\u002Fmbluostudio.com\u002Ftools\u002Fcherry-studio",[494],[53,981,982,983],{},"Cursor IDE 博客 — Cherry Studio 完全指南（2025-03）",[463,984,985],{"href":985,"rel":986},"https:\u002F\u002Fwww.cursor-ide.com\u002Fblog\u002Fcherry-studio-guide",[494],{"title":510,"searchDepth":511,"depth":511,"links":988},[989,990,991,992,993,994,995,996,997,998],{"id":32,"depth":514,"text":33},{"id":48,"depth":514,"text":48},{"id":119,"depth":514,"text":119},{"id":686,"depth":514,"text":687},{"id":189,"depth":514,"text":189},{"id":219,"depth":514,"text":219},{"id":386,"depth":514,"text":386},{"id":427,"depth":514,"text":428},{"id":457,"depth":514,"text":457},{"id":484,"depth":514,"text":484},"\u002Fimg\u002Ftools\u002Fcherry-studio.webp","Cherry Studio 真实评测：开源跨平台桌面 AI 客户端，集成 OpenAI \u002F Claude \u002F Gemini \u002F DeepSeek + Ollama \u002F LM Studio 本地模型，内置 300+ 助手模板 + 本地 RAG 知识库。AGPL-3.0 开源、GitHub 60k+ stars，企业版另询。",[1002,1005,1008,1011],{"q":1003,"a":1004},"Cherry Studio 真的免费吗？","是。客户端完全免费、AGPL-3.0 开源，模型调用走你自己的 API Key（OpenAI \u002F Claude \u002F Gemini \u002F DeepSeek 等付费）或本地 Ollama \u002F LM Studio（零成本）。",{"q":1006,"a":1007},"本地知识库怎么用？","在『知识库』面板新建，拖文件 \u002F 加网址 \u002F 填 sitemap，系统自动向量化（默认 BAAI\u002Fbge-m3 或硅基流动的 Pro 版）；提问时勾选要检索的知识库，AI 会基于检索片段答题并标出来源。",{"q":1009,"a":1010},"和 LobeChat 怎么选？","都开源、多模型、有 RAG。LobeChat 是 Web + 桌面双形态，可自托管 Docker，72k stars；Cherry Studio 是纯桌面（Win\u002FMac\u002FLinux\u002FAndroid），不支持 Web 部署但桌面体验更精细，60k+ stars。要 Web 访问 \u002F 公司多人共享选 LobeChat；个人重度选 Cherry Studio。",{"q":1012,"a":1013},"支持 MCP \u002F 插件吗？","支持 MCP（Model Context Protocol）扩展，配合自定义助手（System Prompt）可扩展工具调用、联网搜索等能力。",[544,545],{},[552,553,554,1017],"android",[1019,1022],{"plan":670,"price":558,"features":1020,"notes":1021},"300+ 助手模板 \u002F 云端 + 本地模型 \u002F 知识库 \u002F MCP \u002F WebDAV 备份","AGPL-3.0 开源",{"plan":676,"price":1023,"features":1024,"notes":1025},"联系销售","私有化部署 \u002F 团队协作 \u002F AI 资源管控 \u002F 知识库管理","面向企业团队","开源免费 \u002F 企业版联系销售",[1028,1029],{"name":569,"url":480},{"name":1030,"url":1031},"Cursor MCP 深度集成","\u002Fplaybook\u002Fonboarding\u002Fcursor-mcp-deep-integration",{"power":574,"ux":525,"price":525,"cn_support":525,"stability":574},{"title":237,"description":1000},"coding\u002Flocal\u002Fcherry-studio",[1036,1038,1040],{"name":1037,"url":971,"accessed":580},"Cherry Studio 官网",{"name":1039,"url":978,"accessed":580},"MBLUO Studio — Cherry Studio 评测",{"name":1041,"url":985,"accessed":580},"Cursor IDE 博客 — Cherry Studio 指南","tools\u002Fcoding\u002Flocal\u002Fcherry-studio","全能 AI 客户端：多模型聚合 + 本地知识库 + 300+ 助手模板，跨平台桌面应用",[524,588,589,1045,590,593,1046],"knowledge-base","china","国产 AI 桌面客户端第一梯队，多模型聚合 + 本地 RAG + 中文体验顶级。需要 Web 部署 \u002F 自托管选 LobeChat；只要桌面体验完整选 Cherry Studio。","https:\u002F\u002Fcherry-ai.com","iH3iDqpTojLLBznROeYXtARpOLDs1yflKA9EMftZkTk",{"id":1051,"title":243,"alternatives":1052,"api_compatible":25,"body":1057,"category":524,"chinese_friendly":511,"cover":1483,"description":1484,"domestic":528,"extension":529,"faq":1485,"free":528,"github":25,"languages":1498,"meta":1499,"models":25,"navigation":547,"notSuitable":25,"opensource":528,"path":24,"pillar":549,"platforms":1500,"priceTable":1501,"pricing":1509,"published":566,"relatedPlaybooks":1510,"relatedReviews":25,"score":1513,"self_host":547,"seo":1514,"slug":1515,"sources":1516,"stem":1523,"suitable":25,"tagline":1524,"tags":1525,"updated":580,"verdict":1532,"website":1533,"__hash__":1534},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Flm-studio.md",[1053,1054,1055,1056],{"name":20,"url":21},{"name":17,"url":18},{"name":14,"url":15},{"name":605,"url":548},{"type":27,"value":1058,"toc":1471},[1059,1061,1068,1071,1073,1130,1132,1146,1151,1155,1159,1176,1180,1197,1199,1225,1227,1366,1368,1400,1402,1425,1427,1446,1448],[30,1060,33],{"id":32},[35,1062,1063,1064,1067],{},"LM Studio 是 Windows \u002F macOS \u002F Linux 桌面应用，让你像浏览 App Store 一样发现、下载、运行本地大模型（GGUF \u002F MLX 格式）。底层基于 llama.cpp + MLX，Mac M 系列原生优化。0.3+ 起新增 Headless 模式 + ",[113,1065,1066],{},"lms"," CLI，可在服务器跑 OpenAI 兼容 API（默认 :1234）。个人 \u002F 评估完全免费，商用咨询。",[35,1069,1070],{},"适合：本地 LLM 入门 \u002F 评估、Mac 用户、需要 GUI 调参 \u002F 模型比较、想给 IDE \u002F 应用接本地 OpenAI 兼容 endpoint 的开发者。不适合：多用户并发生产服务（用 vLLM）、嵌入式 \u002F 边缘部署（用 llama.cpp）、纯 CLI 工作流（用 Ollama）。",[30,1072,48],{"id":48},[50,1074,1075,1081,1087,1093,1103,1112,1118,1124],{},[53,1076,1077,1080],{},[39,1078,1079],{},"模型浏览器","：内置 Hugging Face 检索，按 GGUF \u002F MLX \u002F 大小筛选、一键下载",[53,1082,1083,1086],{},[39,1084,1085],{},"聊天界面","：System Prompt \u002F temperature \u002F top-p \u002F context size 可视化调参",[53,1088,1089,1092],{},[39,1090,1091],{},"多模型并存 \u002F 切换","：同时加载多模型在不同会话中比较",[53,1094,1095,1098,1099,1102],{},[39,1096,1097],{},"OpenAI 兼容 Local Server","：",[113,1100,1101],{},"http:\u002F\u002Flocalhost:1234\u002Fv1","，任何 SDK 即接即用",[53,1104,1105,1098,1108,1111],{},[39,1106,1107],{},"Headless \u002F CLI",[113,1109,1110],{},"lms server start --port 1234","，无 GUI 可跑",[53,1113,1114,1117],{},[39,1115,1116],{},"PDF \u002F 文档对话","：内置基础 RAG，丢文件就能聊",[53,1119,1120,1123],{},[39,1121,1122],{},"MLX 原生支持（Mac）","：M1+ 上比 GGUF + Metal 快 30–50%",[53,1125,1126,1129],{},[39,1127,1128],{},"持续批处理","：Codersera 2026 测得 50–90 tok\u002Fs（消费级 GPU + 中等模型）",[30,1131,119],{"id":119},[50,1133,1134,1140],{},[53,1135,1136,1139],{},[39,1137,1138],{},"个人 \u002F 评估","：免费，全功能可用",[53,1141,1142,1145],{},[39,1143,1144],{},"商用","：邮件 \u002F 官网联系 LM Studio 团队",[679,1147,1148],{},[35,1149,1150],{},"模型本身免费（开源权重），LM Studio 不抽水任何 token 费用。",[30,1152,1154],{"id":1153},"实测mac-m2-pro-qwen3-coder-7b-gguf-q4_k_m","实测（Mac M2 Pro + Qwen3-Coder-7B GGUF Q4_K_M）",[35,1156,1157],{},[39,1158,144],{},[50,1160,1161,1164,1167,1170,1173],{},[53,1162,1163],{},"模型浏览器极舒服：搜「qwen3-coder」直接列出 GGUF + MLX 各 quant，标硬件兼容度",[53,1165,1166],{},"加载 7B Q4 模型 \u003C 3 秒，生成 ~75 tok\u002Fs",[53,1168,1169],{},"Local Server 开了 Cursor 直接接 baseURL → 本地代码补全零成本",[53,1171,1172],{},"MLX 版同模型 ~110 tok\u002Fs，差距显著",[53,1174,1175],{},"多窗口加载 2 个模型并排测，调 prompt 直观",[35,1177,1178],{},[39,1179,169],{},[50,1181,1182,1185,1188,1191,1194],{},[53,1183,1184],{},"模型库依赖 Hugging Face，国内访问要镜像 \u002F 代理",[53,1186,1187],{},"GPU 显存吃满后会自动 offload 到 CPU，无提示就慢下来",[53,1189,1190],{},"Headless 模式相对 Ollama 偏新，文档稍少",[53,1192,1193],{},"闭源应用（虽免费），不适合企业合规挂钩",[53,1195,1196],{},"中文 UI 可用但部分菜单仍英文",[30,1198,189],{"id":189},[191,1200,1201,1204,1207,1210,1213,1220],{},[53,1202,1203],{},"lmstudio.ai 下载（Mac \u002F Windows \u002F Linux）",[53,1205,1206],{},"打开 → Discover 标签 → 搜模型（如 qwen3-coder、deepseek-v3 GGUF\u002FMLX）→ Download",[53,1208,1209],{},"Chat 标签 → 选模型 → 调参聊天",[53,1211,1212],{},"Local Server 标签 → Start Server → 默认端口 1234",[53,1214,1215,1216,1219],{},"在你的应用里：",[113,1217,1218],{},"baseURL = \"http:\u002F\u002Flocalhost:1234\u002Fv1\"","，API Key 任意",[53,1221,1222,1223],{},"Headless：",[113,1224,1110],{},[30,1226,219],{"id":219},[221,1228,1229,1245],{},[224,1230,1231],{},[227,1232,1233,1235,1237,1240,1242],{},[230,1234,232],{},[230,1236,243],{},[230,1238,1239],{},"Ollama",[230,1241,240],{},[230,1243,1244],{},"llama.cpp",[245,1246,1247,1263,1279,1294,1309,1323,1337,1350],{},[227,1248,1249,1251,1254,1257,1260],{},[250,1250,252],{},[250,1252,1253],{},"GUI + CLI",[250,1255,1256],{},"CLI Daemon",[250,1258,1259],{},"Docker UI",[250,1261,1262],{},"二进制",[227,1264,1265,1268,1271,1274,1276],{},[250,1266,1267],{},"模型浏览",[250,1269,1270],{},"✅ 内置",[250,1272,1273],{},"CLI pull",[250,1275,346],{},[250,1277,1278],{},"手动",[227,1280,1281,1284,1286,1289,1292],{},[250,1282,1283],{},"参数调优 GUI",[250,1285,272],{},[250,1287,1288],{},"❌",[250,1290,1291],{},"部分",[250,1293,1288],{},[227,1295,1296,1299,1302,1305,1307],{},[250,1297,1298],{},"OpenAI 兼容 API",[250,1300,1301],{},"✅ :1234",[250,1303,1304],{},"✅ :11434",[250,1306,272],{},[250,1308,272],{},[227,1310,1311,1314,1316,1319,1321],{},[250,1312,1313],{},"MLX (Mac)",[250,1315,272],{},[250,1317,1318],{},"✅ 0.19+",[250,1320,367],{},[250,1322,367],{},[227,1324,1325,1328,1330,1332,1334],{},[250,1326,1327],{},"多用户并发",[250,1329,289],{},[250,1331,289],{},[250,1333,272],{},[250,1335,1336],{},"中",[227,1338,1339,1342,1344,1346,1348],{},[250,1340,1341],{},"开源",[250,1343,383],{},[250,1345,375],{},[250,1347,375],{},[250,1349,375],{},[227,1351,1352,1355,1358,1361,1363],{},[250,1353,1354],{},"上手难度",[250,1356,1357],{},"极低",[250,1359,1360],{},"低",[250,1362,1336],{},[250,1364,1365],{},"高",[30,1367,386],{"id":386},[50,1369,1370,1376,1382,1388,1394],{},[53,1371,1372,1375],{},[39,1373,1374],{},"国内下模型走镜像","：HF 直连慢 \u002F 卡，配 HF_ENDPOINT=hf-mirror.com",[53,1377,1378,1381],{},[39,1379,1380],{},"显存爆 ≠ 报错","：GPU 装不下会无声 offload 到 CPU，关注生成速度，必要时降 quant 或换小模型",[53,1383,1384,1387],{},[39,1385,1386],{},"MLX 优先（Mac M 系列）","：能下 MLX 版就别下 GGUF，速度差距明显",[53,1389,1390,1393],{},[39,1391,1392],{},"Local Server 暴露要谨慎","：默认 0.0.0.0 + 无鉴权，对外开放前加反代 + Bearer",[53,1395,1396,1399],{},[39,1397,1398],{},"闭源合规要核","：企业内部使用前查 license；商用必须联系官方",[30,1401,428],{"id":427},[50,1403,1404,1407,1410,1413,1416,1419,1422],{},[53,1405,1406],{},"✅ 本地 LLM 入门 \u002F 评估",[53,1408,1409],{},"✅ Mac M 系列用户",[53,1411,1412],{},"✅ 想给 Cursor \u002F Cline 接本地 OpenAI 兼容 endpoint",[53,1414,1415],{},"✅ 需要 GUI 调参 \u002F 模型比较",[53,1417,1418],{},"❌ 多用户并发生产服务",[53,1420,1421],{},"❌ 嵌入式 \u002F 边缘设备",[53,1423,1424],{},"❌ 强合规 \u002F 必须开源审计",[30,1426,457],{"id":457},[50,1428,1429,1433,1437,1441],{},[53,1430,1431],{},[463,1432,475],{"href":21},[53,1434,1435],{},[463,1436,470],{"href":18},[53,1438,1439],{},[463,1440,465],{"href":15},[53,1442,1443],{},[463,1444,1445],{"href":572},"Claude Code 上手 Playbook",[30,1447,484],{"id":484},[191,1449,1450,1457,1464],{},[53,1451,1452,1453],{},"LM Studio 官网 ",[463,1454,1455],{"href":1455,"rel":1456},"https:\u002F\u002Flmstudio.ai\u002F",[494],[53,1458,1459,1460],{},"Codersera — LM Studio Complete Guide 2026 ",[463,1461,1462],{"href":1462,"rel":1463},"https:\u002F\u002Fcodersera.com\u002Fblog\u002Flm-studio-complete-guide-2026\u002F",[494],[53,1465,1466,1467],{},"Codersera — Ollama vs LM Studio vs vLLM vs llama.cpp vs MLX 2026 ",[463,1468,1469],{"href":1469,"rel":1470},"https:\u002F\u002Fcodersera.com\u002Fblog\u002Follama-vs-lm-studio-vs-vllm-vs-llama-cpp-vs-mlx-2026\u002F",[494],{"title":510,"searchDepth":511,"depth":511,"links":1472},[1473,1474,1475,1476,1477,1478,1479,1480,1481,1482],{"id":32,"depth":514,"text":33},{"id":48,"depth":514,"text":48},{"id":119,"depth":514,"text":119},{"id":1153,"depth":514,"text":1154},{"id":189,"depth":514,"text":189},{"id":219,"depth":514,"text":219},{"id":386,"depth":514,"text":386},{"id":427,"depth":514,"text":428},{"id":457,"depth":514,"text":457},{"id":484,"depth":514,"text":484},"\u002Fimg\u002Ftools\u002Flm-studio.webp","LM Studio 真实评测：跨平台桌面应用，运行本地 GGUF \u002F MLX 大模型。50–90 tok\u002Fs 持续批处理、OpenAI 兼容本地 API（默认端口 1234）、Headless 模式、Mac \u002F Win 双端。对个人开发者免费，企业咨询。",[1486,1489,1492,1495],{"q":1487,"a":1488},"和 Ollama 怎么选？","LM Studio 是 GUI 优先（模型浏览器 + 参数面板 + 聊天界面），适合个人 \u002F 评估 \u002F 上手。Ollama 是 CLI \u002F Daemon 优先（后台跑 + REST API），适合应用嵌入 \u002F 脚本调用。两者都基于 llama.cpp，在 Mac M 系列上都已用 MLX。",{"q":1490,"a":1491},"支持 MLX 吗？","支持。Mac M1+ 上可加载 MLX 格式模型，速度比 GGUF + Metal 快 30–50%。模型搜索时筛选 MLX 即可。",{"q":1493,"a":1494},"OpenAI 兼容 API 怎么用？","开 Local Server → 默认端口 1234 → `http:\u002F\u002Flocalhost:1234\u002Fv1`。任何 OpenAI SDK 把 baseURL 改这个就能跑本地模型，零代码改动。",{"q":1496,"a":1497},"Headless 模式？","0.3+ 起支持 `lms server start` CLI 启动后台服务，无 GUI 即可跑 OpenAI 兼容 API，适合服务器 \u002F SSH 场景。",[545,544],{},[552,553,554],[1502,1505],{"plan":1138,"price":558,"features":1503,"notes":1504},"全功能 GUI + Headless API + GGUF\u002FMLX","供个人 \u002F 评估使用",{"plan":1144,"price":1506,"features":1507,"notes":1508},"联系咨询","团队部署 \u002F 商用 license","邮件 \u002F 官网联系","免费（个人 \u002F 评估） \u002F 企业 \u002F 商用咨询",[1511,1512],{"name":569,"url":480},{"name":571,"url":572},{"power":574,"ux":525,"price":525,"cn_support":511,"stability":574},{"title":243,"description":1484},"coding\u002Flocal\u002Flm-studio",[1517,1519,1521],{"name":1518,"url":1455,"accessed":580},"LM Studio 官网",{"name":1520,"url":1462,"accessed":580},"Codersera — LM Studio Complete Guide 2026",{"name":1522,"url":1469,"accessed":580},"Codersera — Ollama vs LM Studio vs vLLM 2026","tools\u002Fcoding\u002Flocal\u002Flm-studio","本地 LLM 的 GUI 首选——模型浏览器 + GGUF\u002FMLX 推理 + OpenAI 兼容 API + Mac 原生优化",[524,1526,1527,1528,1529,1530,1531],"gui","gguf","mlx","llama-cpp","mac","openai-compatible","Mac \u002F Windows 桌面本地 LLM 的 GUI 首选——上手最快、模型浏览最舒服、自带 OpenAI 兼容 API。批量服务 \u002F 多用户场景用 vLLM；纯 CLI \u002F 嵌入应用走 Ollama。","https:\u002F\u002Flmstudio.ai","mFDr4hC-XSmdJmmz0qk02JMEKIemqp0HnyhGIxnoWo0",{"id":10,"title":11,"alternatives":1536,"api_compatible":25,"body":1541,"category":524,"chinese_friendly":525,"cover":526,"description":527,"domestic":528,"extension":529,"faq":1894,"free":528,"github":25,"languages":1899,"meta":1900,"models":25,"navigation":547,"notSuitable":25,"opensource":547,"path":548,"pillar":549,"platforms":1901,"priceTable":1902,"pricing":565,"published":566,"relatedPlaybooks":1905,"relatedReviews":25,"score":1908,"self_host":547,"seo":1909,"slug":576,"sources":1910,"stem":585,"suitable":25,"tagline":586,"tags":1914,"updated":580,"verdict":594,"website":595,"__hash__":596},[1537,1538,1539,1540],{"name":14,"url":15},{"name":17,"url":18},{"name":20,"url":21},{"name":23,"url":24},{"type":27,"value":1542,"toc":1882},[1543,1545,1549,1551,1553,1597,1599,1609,1611,1613,1617,1631,1635,1647,1649,1667,1669,1795,1797,1823,1825,1843,1845,1863,1865],[30,1544,33],{"id":32},[35,1546,37,1547,42],{},[39,1548,41],{},[35,1550,45],{},[30,1552,48],{"id":48},[50,1554,1555,1559,1563,1567,1571,1575,1579,1583,1587,1591],{},[53,1556,1557,58],{},[39,1558,57],{},[53,1560,1561,64],{},[39,1562,63],{},[53,1564,1565,70],{},[39,1566,69],{},[53,1568,1569,76],{},[39,1570,75],{},[53,1572,1573,82],{},[39,1574,81],{},[53,1576,1577,88],{},[39,1578,87],{},[53,1580,1581],{},[39,1582,93],{},[53,1584,1585,99],{},[39,1586,98],{},[53,1588,1589,105],{},[39,1590,104],{},[53,1592,1593,111,1595,116],{},[39,1594,110],{},[113,1596,115],{},[30,1598,119],{"id":119},[50,1600,1601,1605],{},[53,1602,1603,127],{},[39,1604,126],{},[53,1606,1607,132],{},[39,1608,104],{},[35,1610,135],{},[30,1612,139],{"id":138},[35,1614,1615],{},[39,1616,144],{},[50,1618,1619,1621,1623,1625,1627,1629],{},[53,1620,149],{},[53,1622,152],{},[53,1624,155],{},[53,1626,158],{},[53,1628,161],{},[53,1630,164],{},[35,1632,1633],{},[39,1634,169],{},[50,1636,1637,1639,1641,1643,1645],{},[53,1638,174],{},[53,1640,177],{},[53,1642,180],{},[53,1644,183],{},[53,1646,186],{},[30,1648,189],{"id":189},[191,1650,1651,1653,1657,1659,1661,1663,1665],{},[53,1652,195],{},[53,1654,198,1655],{},[113,1656,201],{},[53,1658,204],{},[53,1660,207],{},[53,1662,210],{},[53,1664,213],{},[53,1666,216],{},[30,1668,219],{"id":219},[221,1670,1671,1685],{},[224,1672,1673],{},[227,1674,1675,1677,1679,1681,1683],{},[230,1676,232],{},[230,1678,11],{},[230,1680,237],{},[230,1682,240],{},[230,1684,243],{},[245,1686,1687,1699,1711,1723,1735,1747,1759,1771,1783],{},[227,1688,1689,1691,1693,1695,1697],{},[250,1690,252],{},[250,1692,98],{},[250,1694,257],{},[250,1696,260],{},[250,1698,257],{},[227,1700,1701,1703,1705,1707,1709],{},[250,1702,57],{},[250,1704,269],{},[250,1706,272],{},[250,1708,272],{},[250,1710,277],{},[227,1712,1713,1715,1717,1719,1721],{},[250,1714,63],{},[250,1716,284],{},[250,1718,272],{},[250,1720,289],{},[250,1722,289],{},[227,1724,1725,1727,1729,1731,1733],{},[250,1726,296],{},[250,1728,272],{},[250,1730,272],{},[250,1732,303],{},[250,1734,289],{},[227,1736,1737,1739,1741,1743,1745],{},[250,1738,310],{},[250,1740,313],{},[250,1742,316],{},[250,1744,319],{},[250,1746,289],{},[227,1748,1749,1751,1753,1755,1757],{},[250,1750,326],{},[250,1752,272],{},[250,1754,272],{},[250,1756,333],{},[250,1758,289],{},[227,1760,1761,1763,1765,1767,1769],{},[250,1762,340],{},[250,1764,343],{},[250,1766,346],{},[250,1768,284],{},[250,1770,346],{},[227,1772,1773,1775,1777,1779,1781],{},[250,1774,355],{},[250,1776,358],{},[250,1778,361],{},[250,1780,364],{},[250,1782,367],{},[227,1784,1785,1787,1789,1791,1793],{},[250,1786,372],{},[250,1788,375],{},[250,1790,378],{},[250,1792,375],{},[250,1794,383],{},[30,1796,386],{"id":386},[50,1798,1799,1803,1807,1811,1815,1819],{},[53,1800,1801,394],{},[39,1802,393],{},[53,1804,1805,400],{},[39,1806,399],{},[53,1808,1809,406],{},[39,1810,405],{},[53,1812,1813,412],{},[39,1814,411],{},[53,1816,1817,418],{},[39,1818,417],{},[53,1820,1821,424],{},[39,1822,423],{},[30,1824,428],{"id":427},[50,1826,1827,1829,1831,1833,1835,1837,1839,1841],{},[53,1828,433],{},[53,1830,436],{},[53,1832,439],{},[53,1834,442],{},[53,1836,445],{},[53,1838,448],{},[53,1840,451],{},[53,1842,454],{},[30,1844,457],{"id":457},[50,1846,1847,1851,1855,1859],{},[53,1848,1849],{},[463,1850,465],{"href":15},[53,1852,1853],{},[463,1854,470],{"href":18},[53,1856,1857],{},[463,1858,475],{"href":21},[53,1860,1861],{},[463,1862,481],{"href":480},[30,1864,484],{"id":484},[191,1866,1867,1872,1877],{},[53,1868,489,1869],{},[463,1870,492],{"href":492,"rel":1871},[494],[53,1873,497,1874],{},[463,1875,500],{"href":500,"rel":1876},[494],[53,1878,504,1879],{},[463,1880,507],{"href":507,"rel":1881},[494],{"title":510,"searchDepth":511,"depth":511,"links":1883},[1884,1885,1886,1887,1888,1889,1890,1891,1892,1893],{"id":32,"depth":514,"text":33},{"id":48,"depth":514,"text":48},{"id":119,"depth":514,"text":119},{"id":138,"depth":514,"text":139},{"id":189,"depth":514,"text":189},{"id":219,"depth":514,"text":219},{"id":386,"depth":514,"text":386},{"id":427,"depth":514,"text":428},{"id":457,"depth":514,"text":457},{"id":484,"depth":514,"text":484},[1895,1896,1897,1898],{"q":532,"a":533},{"q":535,"a":536},{"q":538,"a":539},{"q":541,"a":542},[544,545],{},[551,552,553,554,555],[1903,1904],{"plan":126,"price":558,"features":559,"notes":560},{"plan":104,"price":562,"features":563,"notes":564},[1906,1907],{"name":569,"url":480},{"name":571,"url":572},{"power":525,"ux":525,"price":525,"cn_support":525,"stability":574},{"title":11,"description":527},[1911,1912,1913],{"name":579,"url":492,"accessed":580},{"name":582,"url":500,"accessed":580},{"name":584,"url":507,"accessed":580},[524,551,588,589,590,591,592,593],{"id":1916,"title":1239,"alternatives":1917,"api_compatible":25,"body":1922,"category":524,"chinese_friendly":511,"cover":2341,"description":2342,"domestic":528,"extension":529,"faq":2343,"free":528,"github":25,"languages":2356,"meta":2357,"models":25,"navigation":547,"notSuitable":25,"opensource":547,"path":21,"pillar":549,"platforms":2358,"priceTable":2359,"pricing":2363,"published":566,"relatedPlaybooks":2364,"relatedReviews":25,"score":2367,"self_host":547,"seo":2368,"slug":2369,"sources":2370,"stem":2376,"suitable":25,"tagline":2377,"tags":2378,"updated":580,"verdict":2383,"website":2384,"__hash__":2385},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Follama.md",[1918,1919,1920,1921],{"name":23,"url":24},{"name":17,"url":18},{"name":14,"url":15},{"name":605,"url":548},{"type":27,"value":1923,"toc":2329},[1924,1926,1933,1936,1938,2006,2008,2011,2015,2019,2039,2043,2074,2076,2110,2112,2227,2229,2261,2263,2286,2288,2306,2308],[30,1925,33],{"id":32},[35,1927,1928,1929,1932],{},"Ollama 是本地 LLM 的 Daemon 事实标准——后台跑、暴露 REST API（11434）+ CLI、Modelfile 配置、GGUF 一站式。MIT 开源，跨 Win \u002F Mac \u002F Linux。0.19+ 起 Mac M 系列底层切 MLX 推理。模型库覆盖 Llama \u002F Qwen \u002F DeepSeek \u002F Gemma \u002F Mistral 等主流开源模型，",[113,1930,1931],{},"ollama pull"," 一键拉。",[35,1934,1935],{},"适合：给 Cursor \u002F Cline \u002F Continue \u002F Open WebUI 接本地 OpenAI 兼容 endpoint、个人 \u002F 评估 \u002F 原型、嵌入应用、自动化脚本。不适合：GUI 偏好用户（用 LM Studio）、多用户并发生产服务（用 vLLM）、模型浏览 \u002F 调参界面（用 LM Studio）。",[30,1937,48],{"id":48},[50,1939,1940,1946,1954,1960,1967,1982,1988,1994,2000],{},[53,1941,1942,1945],{},[39,1943,1944],{},"后台 Daemon","：开机自启，应用调用零延迟",[53,1947,1948,1098,1951],{},[39,1949,1950],{},"CLI",[113,1952,1953],{},"ollama pull \u002F run \u002F list \u002F show \u002F create \u002F serve",[53,1955,1956,1959],{},[39,1957,1958],{},"Modelfile","：类 Dockerfile 注册任意 GGUF，配 SYSTEM \u002F PARAMETER \u002F TEMPLATE",[53,1961,1962,1098,1964],{},[39,1963,1298],{},[113,1965,1966],{},"http:\u002F\u002Flocalhost:11434\u002Fv1\u002Fchat\u002Fcompletions",[53,1968,1969,1098,1972,1975,1976,1975,1979],{},[39,1970,1971],{},"原生 API",[113,1973,1974],{},"\u002Fapi\u002Fchat","、",[113,1977,1978],{},"\u002Fapi\u002Fgenerate",[113,1980,1981],{},"\u002Fapi\u002Fembeddings",[53,1983,1984,1987],{},[39,1985,1986],{},"模型库","：官方注册表内置 Llama \u002F Qwen \u002F DeepSeek \u002F Gemma \u002F Mistral \u002F GPT-OSS 等",[53,1989,1990,1993],{},[39,1991,1992],{},"MLX 加速（Mac）","：0.19+ 起 M 系列自动用 MLX",[53,1995,1996,1999],{},[39,1997,1998],{},"量化","：默认 Q4_K_M、支持 Q5 \u002F Q8 \u002F FP16",[53,2001,2002,2005],{},[39,2003,2004],{},"跨平台","：Win \u002F Mac \u002F Linux 安装包，Docker 官方镜像",[30,2007,119],{"id":119},[35,2009,2010],{},"完全免费、MIT 开源、商用免费。",[30,2012,2014],{"id":2013},"实测m2-pro-qwen3-coder-7b-q4","实测（M2 Pro + Qwen3-Coder-7B Q4）",[35,2016,2017],{},[39,2018,144],{},[50,2020,2021,2027,2030,2033,2036],{},[53,2022,2023,2026],{},[113,2024,2025],{},"ollama run qwen3-coder:7b"," 一行起飞，3 秒进交互",[53,2028,2029],{},"REST API 配 Cursor \u002F Cline \u002F Continue 几乎全工具开箱即用",[53,2031,2032],{},"Modelfile 写自定义编码助手（low temperature + system prompt + 16K context）几分钟搞定",[53,2034,2035],{},"多模型并存，按需切换，内存占用合理",[53,2037,2038],{},"Mac M 系列 MLX 后比旧 GGUF 模式快显著",[35,2040,2041],{},[39,2042,169],{},[50,2044,2045,2055,2061,2068,2071],{},[53,2046,2047,2048,2051,2052],{},"默认 ",[113,2049,2050],{},"num_ctx"," 偏小（2048），跑长上下文要在 Modelfile 加 ",[113,2053,2054],{},"PARAMETER num_ctx 16384",[53,2056,2057,2058],{},"模型默认走 0.0.0.0:11434 ↔ Docker 容器互访要 ",[113,2059,2060],{},"--add-host=host.docker.internal:host-gateway",[53,2062,2063,2064,2067],{},"国内 ",[113,2065,2066],{},"ollama.com\u002Flibrary"," 下载偶有慢，可手动 HF 下 GGUF + Modelfile 自建",[53,2069,2070],{},"多用户并发吞吐显著低于 vLLM",[53,2072,2073],{},"没有 GUI，模型浏览 \u002F 参数面板要走 LM Studio \u002F Open WebUI 配合",[30,2075,189],{"id":189},[191,2077,2078,2084,2090,2095,2101,2107],{},[53,2079,2080,2083],{},[113,2081,2082],{},"curl -fsSL https:\u002F\u002Follama.ai\u002Finstall.sh | sh","（Mac \u002F Linux）；Windows winget",[53,2085,2086,2089],{},[113,2087,2088],{},"ollama pull qwen3-coder:7b","（按需换模型）",[53,2091,2092,2094],{},[113,2093,2025],{}," 直接聊",[53,2096,2097,2098],{},"应用接入：baseURL = ",[113,2099,2100],{},"http:\u002F\u002Flocalhost:11434\u002Fv1",[53,2102,2103,2104],{},"自定义：写 Modelfile → ",[113,2105,2106],{},"ollama create my-coder -f Modelfile",[53,2108,2109],{},"进阶：装 Open WebUI 做前端 \u002F 多人共享",[30,2111,219],{"id":219},[221,2113,2114,2129],{},[224,2115,2116],{},[227,2117,2118,2120,2122,2124,2127],{},[230,2119,232],{},[230,2121,1239],{},[230,2123,243],{},[230,2125,2126],{},"vLLM",[230,2128,1244],{},[245,2130,2131,2147,2159,2172,2185,2201,2213],{},[227,2132,2133,2135,2138,2141,2144],{},[250,2134,252],{},[250,2136,2137],{},"CLI + Daemon",[250,2139,2140],{},"GUI + Headless",[250,2142,2143],{},"Python Server",[250,2145,2146],{},"C++ 二进制",[227,2148,2149,2151,2153,2155,2157],{},[250,2150,189],{},[250,2152,1357],{},[250,2154,1357],{},[250,2156,1336],{},[250,2158,1365],{},[227,2160,2161,2163,2165,2168,2170],{},[250,2162,1267],{},[250,2164,1950],{},[250,2166,2167],{},"✅ GUI",[250,2169,346],{},[250,2171,346],{},[227,2173,2174,2177,2179,2181,2183],{},[250,2175,2176],{},"OpenAI 兼容",[250,2178,1304],{},[250,2180,1301],{},[250,2182,272],{},[250,2184,272],{},[227,2186,2187,2190,2193,2196,2199],{},[250,2188,2189],{},"多用户吞吐",[250,2191,2192],{},"弱（~40 tok\u002Fs）",[250,2194,2195],{},"中（50–90）",[250,2197,2198],{},"强（800–12500）",[250,2200,1336],{},[227,2202,2203,2205,2207,2209,2211],{},[250,2204,1313],{},[250,2206,1318],{},[250,2208,272],{},[250,2210,1291],{},[250,2212,367],{},[227,2214,2215,2217,2219,2222,2225],{},[250,2216,1341],{},[250,2218,375],{},[250,2220,2221],{},"闭源",[250,2223,2224],{},"Apache 2.0",[250,2226,375],{},[30,2228,386],{"id":386},[50,2230,2231,2237,2243,2249,2255],{},[53,2232,2233,2236],{},[39,2234,2235],{},"num_ctx 一定要设","：默认 2K 太小，跑代码 \u002F 长文档要 16K+",[53,2238,2239,2242],{},[39,2240,2241],{},"Modelfile 模板别漏 TEMPLATE","：错的 chat template 会让模型输出乱码 \u002F 不停",[53,2244,2245,2248],{},[39,2246,2247],{},"KV cache 爆表 = 速度悬崖","：32B 模型 32K 上下文，KV cache 可能 12+ GB，超显存自动 offload 慢 10×",[53,2250,2251,2254],{},[39,2252,2253],{},"不要 0.0.0.0 直接对公网","：默认无鉴权，对外暴露走反代 + Bearer \u002F mTLS",[53,2256,2257,2260],{},[39,2258,2259],{},"Mac 让它自动用 MLX","：升 0.19+；不要手动强制 GGUF + Metal",[30,2262,428],{"id":427},[50,2264,2265,2268,2271,2274,2277,2280,2283],{},[53,2266,2267],{},"✅ 应用 \u002F IDE 接本地模型（Cursor \u002F Cline \u002F Continue）",[53,2269,2270],{},"✅ 个人 \u002F 评估 \u002F 脚本自动化",[53,2272,2273],{},"✅ Modelfile 自定义系统 prompt + 参数",[53,2275,2276],{},"✅ Mac M 系列 MLX 用户",[53,2278,2279],{},"❌ 多用户并发生产服务（用 vLLM）",[53,2281,2282],{},"❌ GUI 调参 \u002F 模型浏览（配 LM Studio \u002F Open WebUI）",[53,2284,2285],{},"❌ 极致单卡吞吐研究（直接 llama.cpp \u002F vLLM）",[30,2287,457],{"id":457},[50,2289,2290,2294,2298,2302],{},[53,2291,2292],{},[463,2293,953],{"href":24},[53,2295,2296],{},[463,2297,470],{"href":18},[53,2299,2300],{},[463,2301,465],{"href":15},[53,2303,2304],{},[463,2305,481],{"href":480},[30,2307,484],{"id":484},[191,2309,2310,2317,2324],{},[53,2311,2312,2313],{},"Markaicode — Import GGUF Models into Ollama 2026（2026-05-15）",[463,2314,2315],{"href":2315,"rel":2316},"https:\u002F\u002Fmarkaicode.com\u002Fimport-gguf-models-ollama-guide",[494],[53,2318,2319,2320],{},"ComputingForGeeks — Ollama Models Cheat Sheet 2026 ",[463,2321,2322],{"href":2322,"rel":2323},"https:\u002F\u002Fcomputingforgeeks.com\u002Follama-models-cheat-sheet",[494],[53,2325,1466,2326],{},[463,2327,1469],{"href":1469,"rel":2328},[494],{"title":510,"searchDepth":511,"depth":511,"links":2330},[2331,2332,2333,2334,2335,2336,2337,2338,2339,2340],{"id":32,"depth":514,"text":33},{"id":48,"depth":514,"text":48},{"id":119,"depth":514,"text":119},{"id":2013,"depth":514,"text":2014},{"id":189,"depth":514,"text":189},{"id":219,"depth":514,"text":219},{"id":386,"depth":514,"text":386},{"id":427,"depth":514,"text":428},{"id":457,"depth":514,"text":457},{"id":484,"depth":514,"text":484},"\u002Fimg\u002Ftools\u002Follama.webp","Ollama 真实评测：本地 LLM 的事实标准 Daemon，CLI + REST API，模型库 + Modelfile + GGUF 一站式。0.19+ 在 Mac M 系列用 MLX 加速；OpenAI 兼容端点 11434；MIT 开源 + 跨平台。",[2344,2347,2350,2353],{"q":2345,"a":2346},"和 LM Studio 怎么选？","Ollama = Daemon + CLI，开机自启在 11434 端口跑，应用 \u002F IDE 调它最方便。LM Studio = GUI，模型浏览 \u002F 调参 \u002F 聊天体验更好。两者底层都基于 llama.cpp，Mac M 系列上都已切 MLX。",{"q":2348,"a":2349},"Modelfile 是什么？","类 Dockerfile 的模型配置：`FROM .\u002Fxxx.gguf` + PARAMETER \u002F TEMPLATE \u002F SYSTEM。把任意 GGUF 注册成本地模型供调用。`ollama create my-model -f Modelfile`。",{"q":2351,"a":2352},"OpenAI 兼容端点？","`http:\u002F\u002Flocalhost:11434\u002Fv1`。任何 OpenAI SDK 改 baseURL 即用。也可走原生 `\u002Fapi\u002Fchat`、`\u002Fapi\u002Fgenerate`。",{"q":2354,"a":2355},"并发能力？","单用户原型场景顺滑（~40 tok\u002Fs peak），多用户并发明显不如 vLLM（vLLM 的 PagedAttention + 连续批处理高 16–20×）。生产并发选 vLLM。",[545],{},[552,553,554,555],[2360],{"plan":670,"price":558,"features":2361,"notes":2362},"完整 CLI + REST API + Modelfile + 模型库 + MIT 协议","全平台、商用免费","完全免费 + 开源（MIT）",[2365,2366],{"name":569,"url":480},{"name":571,"url":572},{"power":574,"ux":574,"price":525,"cn_support":511,"stability":525},{"title":1239,"description":2342},"coding\u002Flocal\u002Follama",[2371,2373,2375],{"name":2372,"url":2315,"accessed":580},"Markaicode — Import GGUF 2026",{"name":2374,"url":2322,"accessed":580},"ComputingForGeeks — Ollama Cheat Sheet 2026",{"name":1522,"url":1469,"accessed":580},"tools\u002Fcoding\u002Flocal\u002Follama","本地 LLM 的 Daemon——CLI + REST API 后台跑，给 Cursor \u002F Cline \u002F Open WebUI 接本地模型最低门槛",[524,2379,2380,2381,2382,1527,1528,1531,593],"daemon","cli","rest-api","modelfile","本地 LLM 的 Daemon 事实标准，CLI \u002F Modelfile \u002F REST API 三件套配合最广泛。GUI 偏好用户走 LM Studio；多用户并发生产用 vLLM；其他场景几乎默认 Ollama。","https:\u002F\u002Follama.com","V4PvNLB8lbjAzlhHpWFHSyKzvb328rvWx3nckggFlD8",{"id":2387,"title":240,"alternatives":2388,"api_compatible":25,"body":2393,"category":524,"chinese_friendly":574,"cover":2808,"description":2809,"domestic":528,"extension":529,"faq":2810,"free":528,"github":25,"languages":2823,"meta":2824,"models":25,"navigation":547,"notSuitable":25,"opensource":547,"path":18,"pillar":549,"platforms":2825,"priceTable":2827,"pricing":2834,"published":566,"relatedPlaybooks":2835,"relatedReviews":25,"score":2838,"self_host":547,"seo":2839,"slug":2840,"sources":2841,"stem":2848,"suitable":25,"tagline":2849,"tags":2850,"updated":580,"verdict":2853,"website":2854,"__hash__":2855},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Fopen-webui.md",[2389,2390,2391,2392],{"name":605,"url":548},{"name":14,"url":15},{"name":20,"url":21},{"name":23,"url":24},{"type":27,"value":2394,"toc":2796},[2395,2397,2400,2403,2405,2467,2469,2472,2476,2480,2508,2512,2536,2538,2572,2574,2681,2683,2726,2728,2751,2753,2771,2773],[30,2396,33],{"id":32},[35,2398,2399],{},"Open WebUI（原 Ollama WebUI）是 MIT 开源、自托管 AI 平台，最常见用法是 Docker 跑起来给 Ollama 套一个 ChatGPT 风格前端。GitHub 126k+ stars、282M+ Docker pulls，事实上的本地 AI 前端首选。支持任意 OpenAI 兼容后端 + RAG 知识库 + 多用户账号 + 工具调用 + MCP-OpenAPI 代理 + 联网搜索 + 语音 + 图像生成。",[35,2401,2402],{},"适合：团队 \u002F 家庭 \u002F 公司部署一份共享、要 Web 端访问、多用户分账号、SearXNG 联网搜索、Confluence \u002F S3 \u002F GitHub 数据源同步。不适合：单人桌面体验（用 Cherry Studio）、零运维 \u002F 不愿碰 Docker。",[30,2404,48],{"id":48},[50,2406,2407,2413,2419,2425,2431,2437,2443,2449,2455,2461],{},[53,2408,2409,2412],{},[39,2410,2411],{},"多模型后端","：Ollama \u002F OpenAI \u002F vLLM \u002F Anthropic \u002F Groq \u002F LocalAI \u002F 任意 OpenAI 兼容",[53,2414,2415,2418],{},[39,2416,2417],{},"多用户 + RBAC","：注册 \u002F 邀请 \u002F 角色权限 \u002F 工作区隔离",[53,2420,2421,2424],{},[39,2422,2423],{},"RAG 知识库","：上传文档 \u002F 网址 \u002F SearXNG 联网搜索 → 向量化 → 对话引用",[53,2426,2427,2430],{},[39,2428,2429],{},"Tools \u002F Functions","：Python 写函数即扩展（联网 \u002F 计算器 \u002F 自定义 API）",[53,2432,2433,2436],{},[39,2434,2435],{},"mcpo","：MCP-to-OpenAPI 代理，任意 MCP 服务器接进来",[53,2438,2439,2442],{},[39,2440,2441],{},"oikb","：知识库同步本地文件夹 \u002F GitHub \u002F S3 \u002F Confluence 等 40+ 源",[53,2444,2445,2448],{},[39,2446,2447],{},"open-terminal \u002F cptr","：给 AI 真实终端 + 文件 + 沙箱执行",[53,2450,2451,2454],{},[39,2452,2453],{},"图像生成","：Stable Diffusion \u002F DALL·E \u002F 自托管接入",[53,2456,2457,2460],{},[39,2458,2459],{},"语音输入 \u002F TTS","：内置",[53,2462,2463,2466],{},[39,2464,2465],{},"企业 LTS","：custom branding + SLA + 长期支持版本（联系销售）",[30,2468,119],{"id":119},[35,2470,2471],{},"完全免费、MIT 开源、商用免费。Enterprise 提供品牌定制 + SLA + LTS。",[30,2473,2475],{"id":2474},"实测ubuntu-2404-ollama-后端-5-人小团队","实测（Ubuntu 24.04 + Ollama 后端 + 5 人小团队）",[35,2477,2478],{},[39,2479,144],{},[50,2481,2482,2489,2492,2499,2502,2505],{},[53,2483,2484,2485,2488],{},"单条 ",[113,2486,2487],{},"docker run"," 五分钟上线",[53,2490,2491],{},"自带的多用户 + 角色权限省去重新搭 Auth",[53,2493,2494,2495,2498],{},"RAG 直传 30 个 PDF 后向量化顺利，对话中 ",[113,2496,2497],{},"#知识库"," 引用准确",[53,2500,2501],{},"mcpo 把 GitHub MCP 服务器接进来，团队对话里直接 issue \u002F PR 操作",[53,2503,2504],{},"模型切换流畅，OpenAI + Ollama 并存",[53,2506,2507],{},"SearXNG 联网搜索给模型实时信息，过时知识截止问题缓解",[35,2509,2510],{},[39,2511,169],{},[50,2513,2514,2517,2523,2530,2533],{},[53,2515,2516],{},"Docker 镜像 ~1.5GB，首次拉取偏慢",[53,2518,2047,2519,2522],{},[113,2520,2521],{},"0.0.0.0"," 公网暴露要加 HTTPS + 反代",[53,2524,2525,2526,2529],{},"嵌入模型 ",[113,2527,2528],{},"sentence-transformers"," 中文效果一般，建议换 bge-m3",[53,2531,2532],{},"多用户共享 Ollama 时并发吞吐瓶颈在 Ollama，不在 Open WebUI（生产用 vLLM 后端）",[53,2534,2535],{},"版本升级要看 changelog，部分 minor 含 breaking 改动",[30,2537,189],{"id":189},[191,2539,2540,2546,2553,2556,2559,2562,2565],{},[53,2541,2542,2543],{},"装 Docker → ",[113,2544,2545],{},"docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:\u002Fapp\u002Fbackend\u002Fdata --name open-webui --restart always ghcr.io\u002Fopen-webui\u002Fopen-webui:main",[53,2547,2548,2549,2552],{},"浏览器开 ",[113,2550,2551],{},"http:\u002F\u002Flocalhost:3000"," → 注册第一个账号（管理员）",[53,2554,2555],{},"设置 → Connections → 连接 Ollama \u002F 加 OpenAI Key",[53,2557,2558],{},"Models → Pull \u002F Discover 模型",[53,2560,2561],{},"Workspaces → 建知识库 → 上传文档",[53,2563,2564],{},"Tools → 启用 \u002F 写自定义函数",[53,2566,2567,2568,2571],{},"生产部署：Nginx 反代 + Let's Encrypt + 备份 ",[113,2569,2570],{},"\u002Fapp\u002Fbackend\u002Fdata"," volume",[30,2573,219],{"id":219},[221,2575,2576,2590],{},[224,2577,2578],{},[227,2579,2580,2582,2584,2586,2588],{},[230,2581,232],{},[230,2583,240],{},[230,2585,11],{},[230,2587,237],{},[230,2589,243],{},[245,2591,2592,2604,2616,2630,2643,2657,2669],{},[227,2593,2594,2596,2598,2600,2602],{},[250,2595,252],{},[250,2597,260],{},[250,2599,784],{},[250,2601,257],{},[250,2603,257],{},[227,2605,2606,2608,2610,2612,2614],{},[250,2607,340],{},[250,2609,284],{},[250,2611,272],{},[250,2613,346],{},[250,2615,346],{},[227,2617,2618,2621,2624,2626,2628],{},[250,2619,2620],{},"RAG",[250,2622,2623],{},"✅ 强 + oikb",[250,2625,272],{},[250,2627,272],{},[250,2629,289],{},[227,2631,2632,2635,2637,2639,2641],{},[250,2633,2634],{},"工具 \u002F MCP",[250,2636,333],{},[250,2638,272],{},[250,2640,272],{},[250,2642,289],{},[227,2644,2645,2648,2651,2653,2655],{},[250,2646,2647],{},"自托管",[250,2649,2650],{},"✅ Docker \u002F K8s",[250,2652,837],{},[250,2654,346],{},[250,2656,346],{},[227,2658,2659,2661,2663,2665,2667],{},[250,2660,355],{},[250,2662,364],{},[250,2664,358],{},[250,2666,361],{},[250,2668,367],{},[227,2670,2671,2673,2675,2677,2679],{},[250,2672,372],{},[250,2674,375],{},[250,2676,375],{},[250,2678,378],{},[250,2680,383],{},[30,2682,386],{"id":386},[50,2684,2685,2691,2699,2708,2714,2720],{},[53,2686,2687,2690],{},[39,2688,2689],{},"不要裸 0.0.0.0 + HTTP 暴露公网","：默认无 HTTPS，必上反代 + 强密码 + 速率限制",[53,2692,2693,2698],{},[39,2694,2695,2696,2571],{},"备份 ",[113,2697,2570],{},"：知识库 \u002F 用户 \u002F 对话全在里面",[53,2700,2701,2704,2705,2707],{},[39,2702,2703],{},"中文 RAG 换嵌入模型","：默认 ",[113,2706,2528],{}," 中文一般，配 bge-m3 或硅基流动嵌入 API",[53,2709,2710,2713],{},[39,2711,2712],{},"mcpo 工具范围谨慎","：MCP 给 AI 真实能力，第三方服务器审一遍",[53,2715,2716,2719],{},[39,2717,2718],{},"后端吞吐看 Ollama","：5+ 并发上 vLLM 后端，Ollama 单 worker 会排队",[53,2721,2722,2725],{},[39,2723,2724],{},"升级前看 changelog","：weekly 更新，偶有 breaking",[30,2727,428],{"id":427},[50,2729,2730,2733,2736,2739,2742,2745,2748],{},[53,2731,2732],{},"✅ 团队 \u002F 家庭 \u002F 公司多人共享 AI 平台",[53,2734,2735],{},"✅ 要 Web 端访问 \u002F 移动端兼容",[53,2737,2738],{},"✅ 自托管 \u002F 完全控制数据",[53,2740,2741],{},"✅ MCP \u002F 工具调用刚需",[53,2743,2744],{},"❌ 单人桌面体验（用 Cherry Studio）",[53,2746,2747],{},"❌ 零运维 \u002F 不愿碰 Docker",[53,2749,2750],{},"❌ iOS 原生 App 主力",[30,2752,457],{"id":457},[50,2754,2755,2759,2763,2767],{},[53,2756,2757],{},[463,2758,948],{"href":548},[53,2760,2761],{},[463,2762,465],{"href":15},[53,2764,2765],{},[463,2766,475],{"href":21},[53,2768,2769],{},[463,2770,481],{"href":480},[30,2772,484],{"id":484},[191,2774,2775,2782,2789],{},[53,2776,2777,2778],{},"Open WebUI 官方文档 ",[463,2779,2780],{"href":2780,"rel":2781},"https:\u002F\u002Fdocs.openwebui.com\u002F",[494],[53,2783,2784,2785],{},"Local AI Master — Open WebUI Setup Guide 2026 ",[463,2786,2787],{"href":2787,"rel":2788},"https:\u002F\u002Flocalaimaster.com\u002Fblog\u002Fopen-webui-setup-guide",[494],[53,2790,2791,2792],{},"AIToolDiscovery — Set Up Open-WebUI with Ollama 2026 ",[463,2793,2794],{"href":2794,"rel":2795},"https:\u002F\u002Fwww.aitooldiscovery.com\u002Fhow-to\u002Fsetup-open-webui-ollama",[494],{"title":510,"searchDepth":511,"depth":511,"links":2797},[2798,2799,2800,2801,2802,2803,2804,2805,2806,2807],{"id":32,"depth":514,"text":33},{"id":48,"depth":514,"text":48},{"id":119,"depth":514,"text":119},{"id":2474,"depth":514,"text":2475},{"id":189,"depth":514,"text":189},{"id":219,"depth":514,"text":219},{"id":386,"depth":514,"text":386},{"id":427,"depth":514,"text":428},{"id":457,"depth":514,"text":457},{"id":484,"depth":514,"text":484},"\u002Fimg\u002Ftools\u002Fopen-webui.webp","Open WebUI 真实评测：MIT 开源、自托管 AI 平台，离线优先。Docker 一行起飞、支持 Ollama \u002F OpenAI \u002F vLLM \u002F Anthropic \u002F Groq 等后端，内置 RAG 知识库 + 多用户 + 联网搜索 + 工具调用。GitHub 126k+ stars，事实标准本地 AI 前端。",[2811,2814,2817,2820],{"q":2812,"a":2813},"Docker 一行命令真的够用吗？","够。`docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:\u002Fapp\u002Fbackend\u002Fdata --name open-webui --restart always ghcr.io\u002Fopen-webui\u002Fopen-webui:main`，5 分钟可上线、能多人注册、能接 Ollama \u002F OpenAI。生产再加反代 + HTTPS + 备份。",{"q":2815,"a":2816},"支持哪些模型后端？","Ollama（首选）+ 任何 OpenAI 兼容 endpoint：OpenAI 官方 \u002F Anthropic（OpenAI 兼容代理）\u002F vLLM \u002F Groq \u002F LocalAI \u002F 自建 baseURL。可同时配多个，对话中切换。",{"q":2818,"a":2819},"RAG \u002F 知识库怎么做？","内置：上传 PDF \u002F DOCX \u002F TXT、网址抓取、SearXNG 联网搜索 → 自动向量化 → 在对话中 `#` 引用知识库。配套 oikb 项目可同步本地文件夹 \u002F GitHub \u002F S3 \u002F Confluence 等 40+ 数据源。",{"q":2821,"a":2822},"MCP 怎么接？","通过 mcpo（官方的 MCP-to-OpenAPI 代理）把任意 MCP 服务器暴露成 OpenAPI 工具，再在 Open WebUI 注册即可。无需写 glue code。",[545,544],{},[555,554,553,552,2826],"kubernetes",[2828,2830],{"plan":670,"price":558,"features":2829,"notes":560},"全功能 \u002F 多用户 \u002F RAG \u002F Tools \u002F 联网搜索 \u002F MCP-OpenAPI 代理 \u002F Docker \u002F K8s",{"plan":676,"price":2831,"features":2832,"notes":2833},"咨询","Custom branding \u002F SLA \u002F LTS 长期支持版本","邮件官方","完全免费（MIT 开源） \u002F Enterprise SLA 联系",[2836,2837],{"name":569,"url":480},{"name":571,"url":572},{"power":525,"ux":574,"price":525,"cn_support":574,"stability":525},{"title":240,"description":2809},"coding\u002Flocal\u002Fopen-webui",[2842,2844,2846],{"name":2843,"url":2780,"accessed":580},"Open WebUI 官方文档",{"name":2845,"url":2787,"accessed":580},"Local AI Master — Open WebUI Setup Guide 2026",{"name":2847,"url":2794,"accessed":580},"AIToolDiscovery — Open-WebUI with Ollama 2026","tools\u002Fcoding\u002Flocal\u002Fopen-webui","自托管的 ChatGPT 替代——Ollama \u002F OpenAI 兼容、多用户、RAG、126k+ GitHub stars",[524,2851,555,590,2852,20,593],"self-host","multi-user","自托管多用户 AI 前端的事实标准。团队 \u002F 家庭 \u002F 公司部署一份共享，多模型聚合 + RAG + 工具调用全有。单机 \u002F 桌面体验首选 Cherry Studio \u002F LobeChat。","https:\u002F\u002Fdocs.openwebui.com","8iXO-BuvdCKJCMIXxc6Jlp_MhDAMVBwYJJnhXo6dErw",[],1782316490682]