[{"data":1,"prerenderedAt":2473},["ShallowReactive",2],{"tool-tools\u002Fcoding\u002Flocal":3,"header-counts":4,"footer-counts":7,"tools-coding-local":10},null,{"tools":5,"reviews":6},65,7,{"tools":5,"reviews":6,"playbooks":8,"news":9},10,8,[11,548,1039,1532,2003],{"id":12,"title":13,"alternatives":14,"api_compatible":3,"body":27,"category":476,"chinese_friendly":477,"cover":478,"description":479,"domestic":480,"extension":481,"faq":482,"free":480,"github":3,"languages":495,"meta":498,"models":3,"navigation":499,"notSuitable":3,"opensource":499,"path":500,"pillar":501,"platforms":502,"priceTable":507,"pricing":516,"published":517,"relatedPlaybooks":518,"relatedReviews":3,"score":524,"self_host":499,"seo":526,"slug":527,"sources":528,"stem":536,"suitable":3,"tagline":537,"tags":538,"updated":531,"verdict":545,"website":546,"__hash__":547},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Fcherry-studio.md","Cherry Studio",[15,18,21,24],{"name":16,"url":17},"lobe-chat","\u002Ftools\u002Fcoding\u002Flocal\u002Flobe-chat",{"name":19,"url":20},"lm-studio","\u002Ftools\u002Fcoding\u002Flocal\u002Flm-studio",{"name":22,"url":23},"ollama","\u002Ftools\u002Fcoding\u002Flocal\u002Follama",{"name":25,"url":26},"open-webui","\u002Ftools\u002Fcoding\u002Flocal\u002Fopen-webui",{"type":28,"value":29,"toc":461},"minimark",[30,35,39,42,45,92,95,109,115,119,124,141,146,163,166,193,196,347,350,382,386,409,412,435,438],[31,32,34],"h2",{"id":33},"tldr","TL;DR",[36,37,38],"p",{},"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。",[36,40,41],{},"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。",[31,43,44],{"id":44},"核心能力",[46,47,48,56,62,68,74,80,86],"ul",{},[49,50,51,55],"li",{},[52,53,54],"strong",{},"多模型聚合","：OpenAI \u002F Claude \u002F Gemini \u002F DeepSeek \u002F Qwen \u002F Kimi \u002F Moonshot 等云端 + Ollama \u002F LM Studio 本地",[49,57,58,61],{},[52,59,60],{},"本地 RAG 知识库","：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",[49,63,64,67],{},[52,65,66],{},"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",[49,69,70,73],{},[52,71,72],{},"MCP 协议","：扩展工具调用 \u002F 联网搜索 \u002F 文件操作",[49,75,76,79],{},[52,77,78],{},"数据本地优先","：对话历史本地存储，WebDAV 同步，不上传第三方",[49,81,82,85],{},[52,83,84],{},"多模态","：图片识别 \u002F PDF 阅读 \u002F Markdown + Mermaid + 代码高亮",[49,87,88,91],{},[52,89,90],{},"AI 绘画 + 翻译","：内置主流 SD \u002F DALL·E \u002F 翻译 API 集成",[31,93,94],{"id":94},"价格",[46,96,97,103],{},[49,98,99,102],{},[52,100,101],{},"开源版","：完全免费，AGPL-3.0",[49,104,105,108],{},[52,106,107],{},"Enterprise","：私有化部署 + 团队协作 + 资源管控，联系销售",[110,111,112],"blockquote",{},[36,113,114],{},"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。",[31,116,118],{"id":117},"实测mac-m2-中型知识库","实测（Mac M2 + 中型知识库）",[36,120,121],{},[52,122,123],{},"亮点：",[46,125,126,129,132,135,138],{},[49,127,128],{},"中文 UI \u002F 文档 \u002F 社区都顶级，零门槛上手",[49,130,131],{},"本地 RAG 拖入 30+ PDF 后向量化 \u003C 2 分钟（用 bge-m3）",[49,133,134],{},"多模型并排回答：让 Claude \u002F GPT \u002F DeepSeek 同回一个问题做比较",[49,136,137],{},"MCP 接 Brave Search + 自定义工具流畅",[49,139,140],{},"WebDAV 同步坚果云 \u002F 阿里云盘，桌面 + 移动设备数据互通",[36,142,143],{},[52,144,145],{},"踩坑：",[46,147,148,151,154,157,160],{},[49,149,150],{},"没有 Web 端 \u002F Docker 自托管（要这个用 LobeChat）",[49,152,153],{},"iOS 版尚未发布（roadmap 中）",[49,155,156],{},"大型 PDF（>100 MB）向量化偶有失败，要切小",[49,158,159],{},"助手市场质量参差，要自筛",[49,161,162],{},"模型 API 调用全靠你自己付费，新手要先理解 API Key 概念",[31,164,165],{"id":165},"上手",[167,168,169,172,175,184,187,190],"ol",{},[49,170,171],{},"cherry-ai.com 下载客户端（或 GitHub releases）",[49,173,174],{},"设置 → 模型服务 → 填 OpenAI \u002F Claude \u002F DeepSeek API Key",[49,176,177,178],{},"（可选）本地：装 Ollama → Cherry Studio 自动识别 endpoint ",[179,180,181],"a",{"href":181,"rel":182},"http:\u002F\u002Flocalhost:11434",[183],"nofollow",[49,185,186],{},"新建知识库 → 拖文件 \u002F 加网址 → 等向量化",[49,188,189],{},"新对话 → 选模型 → 勾知识库 → 提问",[49,191,192],{},"进阶：自定义助手（System Prompt）+ MCP 扩展工具",[31,194,195],{"id":195},"对比",[197,198,199,220],"table",{},[200,201,202],"thead",{},[203,204,205,209,211,214,217],"tr",{},[206,207,208],"th",{},"维度",[206,210,13],{},[206,212,213],{},"LobeChat",[206,215,216],{},"LM Studio",[206,218,219],{},"Open WebUI",[221,222,223,240,254,270,283,299,314,330],"tbody",{},[203,224,225,229,232,235,237],{},[226,227,228],"td",{},"形态",[226,230,231],{},"桌面",[226,233,234],{},"Web + 桌面",[226,236,231],{},[226,238,239],{},"Docker \u002F 桌面",[203,241,242,244,247,249,252],{},[226,243,54],{},[226,245,246],{},"✅ 云 + 本地",[226,248,246],{},[226,250,251],{},"本地为主",[226,253,246],{},[203,255,256,259,262,264,267],{},[226,257,258],{},"知识库 RAG",[226,260,261],{},"✅ 强",[226,263,261],{},[226,265,266],{},"弱",[226,268,269],{},"✅",[203,271,272,275,277,279,281],{},[226,273,274],{},"MCP",[226,276,269],{},[226,278,269],{},[226,280,266],{},[226,282,269],{},[203,284,285,288,291,294,297],{},[226,286,287],{},"自托管 \u002F Web",[226,289,290],{},"无 Web",[226,292,293],{},"✅ Docker",[226,295,296],{},"无",[226,298,293],{},[203,300,301,304,307,309,312],{},[226,302,303],{},"中文",[226,305,306],{},"5\u002F5",[226,308,306],{},[226,310,311],{},"4\u002F5",[226,313,311],{},[203,315,316,319,322,325,328],{},[226,317,318],{},"开源协议",[226,320,321],{},"AGPL-3.0",[226,323,324],{},"MIT",[226,326,327],{},"闭源（免费）",[226,329,324],{},[203,331,332,335,338,341,344],{},[226,333,334],{},"GitHub Stars",[226,336,337],{},"60k+",[226,339,340],{},"72k+",[226,342,343],{},"–",[226,345,346],{},"126k+",[31,348,349],{"id":349},"避坑",[46,351,352,358,364,370,376],{},[49,353,354,357],{},[52,355,356],{},"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",[49,359,360,363],{},[52,361,362],{},"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",[49,365,366,369],{},[52,367,368],{},"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",[49,371,372,375],{},[52,373,374],{},"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",[49,377,378,381],{},[52,379,380],{},"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",[31,383,385],{"id":384},"适合-不适合","适合 \u002F 不适合",[46,387,388,391,394,397,400,403,406],{},[49,389,390],{},"✅ 中文用户、AI 重度使用 \u002F 多模型管理",[49,392,393],{},"✅ 需要本地 RAG 知识库",[49,395,396],{},"✅ 关注数据隐私 \u002F 本地存储",[49,398,399],{},"✅ 想用 Ollama \u002F LM Studio 本地模型",[49,401,402],{},"❌ 需要 Web 端 \u002F Docker 自托管",[49,404,405],{},"❌ 团队多人共享 \u002F SSO",[49,407,408],{},"❌ iOS 主力用户",[31,410,411],{"id":411},"相关阅读",[46,413,414,419,424,429],{},[49,415,416],{},[179,417,418],{"href":17},"LobeChat 评测",[49,420,421],{},[179,422,423],{"href":20},"LM Studio 评测",[49,425,426],{},[179,427,428],{"href":23},"Ollama 评测",[49,430,431],{},[179,432,434],{"href":433},"\u002Fplaybook\u002Fonboarding\u002Frag-pipeline-build","RAG Pipeline 搭建 Playbook",[31,436,437],{"id":437},"来源",[167,439,440,447,454],{},[49,441,442,443],{},"Cherry Studio 官网（功能 + 下载）",[179,444,445],{"href":445,"rel":446},"https:\u002F\u002Fwww.cherry-ai.com\u002F",[183],[49,448,449,450],{},"MBLUO Studio — Cherry Studio 评测 2026 ",[179,451,452],{"href":452,"rel":453},"https:\u002F\u002Fmbluostudio.com\u002Ftools\u002Fcherry-studio",[183],[49,455,456,457],{},"Cursor IDE 博客 — Cherry Studio 完全指南（2025-03）",[179,458,459],{"href":459,"rel":460},"https:\u002F\u002Fwww.cursor-ide.com\u002Fblog\u002Fcherry-studio-guide",[183],{"title":462,"searchDepth":463,"depth":463,"links":464},"",3,[465,467,468,469,470,471,472,473,474,475],{"id":33,"depth":466,"text":34},2,{"id":44,"depth":466,"text":44},{"id":94,"depth":466,"text":94},{"id":117,"depth":466,"text":118},{"id":165,"depth":466,"text":165},{"id":195,"depth":466,"text":195},{"id":349,"depth":466,"text":349},{"id":384,"depth":466,"text":385},{"id":411,"depth":466,"text":411},{"id":437,"depth":466,"text":437},"local",5,"\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。",false,"md",[483,486,489,492],{"q":484,"a":485},"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":487,"a":488},"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":490,"a":491},"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":493,"a":494},"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。",[496,497],"zh","en",{},true,"\u002Ftools\u002Fcoding\u002Flocal\u002Fcherry-studio","coding",[503,504,505,506],"windows","macos","linux","android",[508,512],{"plan":101,"price":509,"features":510,"notes":511},"免费","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":107,"price":513,"features":514,"notes":515},"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","2026-06-19",[519,521],{"name":520,"url":433},"RAG Pipeline 搭建",{"name":522,"url":523},"Cursor MCP 深度集成","\u002Fplaybook\u002Fonboarding\u002Fcursor-mcp-deep-integration",{"power":525,"ux":477,"price":477,"cn_support":477,"stability":525},4,{"title":13,"description":479},"coding\u002Flocal\u002Fcherry-studio",[529,532,534],{"name":530,"url":445,"accessed":531},"Cherry Studio 官网","2026-06-24",{"name":533,"url":452,"accessed":531},"MBLUO Studio — Cherry Studio 评测",{"name":535,"url":459,"accessed":531},"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",[476,539,540,541,542,543,544],"desktop","multi-model","knowledge-base","rag","open-source","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":549,"title":216,"alternatives":550,"api_compatible":3,"body":556,"category":476,"chinese_friendly":463,"cover":986,"description":987,"domestic":480,"extension":481,"faq":988,"free":480,"github":3,"languages":1001,"meta":1002,"models":3,"navigation":499,"notSuitable":3,"opensource":480,"path":20,"pillar":501,"platforms":1003,"priceTable":1004,"pricing":1012,"published":517,"relatedPlaybooks":1013,"relatedReviews":3,"score":1017,"self_host":499,"seo":1018,"slug":1019,"sources":1020,"stem":1027,"suitable":3,"tagline":1028,"tags":1029,"updated":531,"verdict":1036,"website":1037,"__hash__":1038},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Flm-studio.md",[551,552,553,555],{"name":22,"url":23},{"name":25,"url":26},{"name":554,"url":500},"cherry-studio",{"name":16,"url":17},{"type":28,"value":557,"toc":974},[558,560,568,571,573,630,632,646,651,655,659,676,680,697,699,725,727,866,868,900,902,925,927,949,951],[31,559,34],{"id":33},[36,561,562,563,567],{},"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 + ",[564,565,566],"code",{},"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。",[36,569,570],{},"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）。",[31,572,44],{"id":44},[46,574,575,581,587,593,603,612,618,624],{},[49,576,577,580],{},[52,578,579],{},"模型浏览器","：内置 Hugging Face 检索，按 GGUF \u002F MLX \u002F 大小筛选、一键下载",[49,582,583,586],{},[52,584,585],{},"聊天界面","：System Prompt \u002F temperature \u002F top-p \u002F context size 可视化调参",[49,588,589,592],{},[52,590,591],{},"多模型并存 \u002F 切换","：同时加载多模型在不同会话中比较",[49,594,595,598,599,602],{},[52,596,597],{},"OpenAI 兼容 Local Server","：",[564,600,601],{},"http:\u002F\u002Flocalhost:1234\u002Fv1","，任何 SDK 即接即用",[49,604,605,598,608,611],{},[52,606,607],{},"Headless \u002F CLI",[564,609,610],{},"lms server start --port 1234","，无 GUI 可跑",[49,613,614,617],{},[52,615,616],{},"PDF \u002F 文档对话","：内置基础 RAG，丢文件就能聊",[49,619,620,623],{},[52,621,622],{},"MLX 原生支持（Mac）","：M1+ 上比 GGUF + Metal 快 30–50%",[49,625,626,629],{},[52,627,628],{},"持续批处理","：Codersera 2026 测得 50–90 tok\u002Fs（消费级 GPU + 中等模型）",[31,631,94],{"id":94},[46,633,634,640],{},[49,635,636,639],{},[52,637,638],{},"个人 \u002F 评估","：免费，全功能可用",[49,641,642,645],{},[52,643,644],{},"商用","：邮件 \u002F 官网联系 LM Studio 团队",[110,647,648],{},[36,649,650],{},"模型本身免费（开源权重），LM Studio 不抽水任何 token 费用。",[31,652,654],{"id":653},"实测mac-m2-pro-qwen3-coder-7b-gguf-q4_k_m","实测（Mac M2 Pro + Qwen3-Coder-7B GGUF Q4_K_M）",[36,656,657],{},[52,658,123],{},[46,660,661,664,667,670,673],{},[49,662,663],{},"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",[49,665,666],{},"加载 7B Q4 模型 \u003C 3 秒，生成 ~75 tok\u002Fs",[49,668,669],{},"Local Server 开了 Cursor 直接接 baseURL → 本地代码补全零成本",[49,671,672],{},"MLX 版同模型 ~110 tok\u002Fs，差距显著",[49,674,675],{},"多窗口加载 2 个模型并排测，调 prompt 直观",[36,677,678],{},[52,679,145],{},[46,681,682,685,688,691,694],{},[49,683,684],{},"模型库依赖 Hugging Face，国内访问要镜像 \u002F 代理",[49,686,687],{},"GPU 显存吃满后会自动 offload 到 CPU，无提示就慢下来",[49,689,690],{},"Headless 模式相对 Ollama 偏新，文档稍少",[49,692,693],{},"闭源应用（虽免费），不适合企业合规挂钩",[49,695,696],{},"中文 UI 可用但部分菜单仍英文",[31,698,165],{"id":165},[167,700,701,704,707,710,713,720],{},[49,702,703],{},"lmstudio.ai 下载（Mac \u002F Windows \u002F Linux）",[49,705,706],{},"打开 → Discover 标签 → 搜模型（如 qwen3-coder、deepseek-v3 GGUF\u002FMLX）→ Download",[49,708,709],{},"Chat 标签 → 选模型 → 调参聊天",[49,711,712],{},"Local Server 标签 → Start Server → 默认端口 1234",[49,714,715,716,719],{},"在你的应用里：",[564,717,718],{},"baseURL = \"http:\u002F\u002Flocalhost:1234\u002Fv1\"","，API Key 任意",[49,721,722,723],{},"Headless：",[564,724,610],{},[31,726,195],{"id":195},[197,728,729,745],{},[200,730,731],{},[203,732,733,735,737,740,742],{},[206,734,208],{},[206,736,216],{},[206,738,739],{},"Ollama",[206,741,219],{},[206,743,744],{},"llama.cpp",[221,746,747,763,779,794,809,823,837,850],{},[203,748,749,751,754,757,760],{},[226,750,228],{},[226,752,753],{},"GUI + CLI",[226,755,756],{},"CLI Daemon",[226,758,759],{},"Docker UI",[226,761,762],{},"二进制",[203,764,765,768,771,774,776],{},[226,766,767],{},"模型浏览",[226,769,770],{},"✅ 内置",[226,772,773],{},"CLI pull",[226,775,296],{},[226,777,778],{},"手动",[203,780,781,784,786,789,792],{},[226,782,783],{},"参数调优 GUI",[226,785,269],{},[226,787,788],{},"❌",[226,790,791],{},"部分",[226,793,788],{},[203,795,796,799,802,805,807],{},[226,797,798],{},"OpenAI 兼容 API",[226,800,801],{},"✅ :1234",[226,803,804],{},"✅ :11434",[226,806,269],{},[226,808,269],{},[203,810,811,814,816,819,821],{},[226,812,813],{},"MLX (Mac)",[226,815,269],{},[226,817,818],{},"✅ 0.19+",[226,820,343],{},[226,822,343],{},[203,824,825,828,830,832,834],{},[226,826,827],{},"多用户并发",[226,829,266],{},[226,831,266],{},[226,833,269],{},[226,835,836],{},"中",[203,838,839,842,844,846,848],{},[226,840,841],{},"开源",[226,843,327],{},[226,845,324],{},[226,847,324],{},[226,849,324],{},[203,851,852,855,858,861,863],{},[226,853,854],{},"上手难度",[226,856,857],{},"极低",[226,859,860],{},"低",[226,862,836],{},[226,864,865],{},"高",[31,867,349],{"id":349},[46,869,870,876,882,888,894],{},[49,871,872,875],{},[52,873,874],{},"国内下模型走镜像","：HF 直连慢 \u002F 卡，配 HF_ENDPOINT=hf-mirror.com",[49,877,878,881],{},[52,879,880],{},"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",[49,883,884,887],{},[52,885,886],{},"MLX 优先（Mac M 系列）","：能下 MLX 版就别下 GGUF，速度差距明显",[49,889,890,893],{},[52,891,892],{},"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",[49,895,896,899],{},[52,897,898],{},"闭源合规要核","：企业内部使用前查 license；商用必须联系官方",[31,901,385],{"id":384},[46,903,904,907,910,913,916,919,922],{},[49,905,906],{},"✅ 本地 LLM 入门 \u002F 评估",[49,908,909],{},"✅ Mac M 系列用户",[49,911,912],{},"✅ 想给 Cursor \u002F Cline 接本地 OpenAI 兼容 endpoint",[49,914,915],{},"✅ 需要 GUI 调参 \u002F 模型比较",[49,917,918],{},"❌ 多用户并发生产服务",[49,920,921],{},"❌ 嵌入式 \u002F 边缘设备",[49,923,924],{},"❌ 强合规 \u002F 必须开源审计",[31,926,411],{"id":411},[46,928,929,933,938,943],{},[49,930,931],{},[179,932,428],{"href":23},[49,934,935],{},[179,936,937],{"href":26},"Open WebUI 评测",[49,939,940],{},[179,941,942],{"href":500},"Cherry Studio 评测",[49,944,945],{},[179,946,948],{"href":947},"\u002Fplaybook\u002Fonboarding\u002Fclaude-code-getting-started","Claude Code 上手 Playbook",[31,950,437],{"id":437},[167,952,953,960,967],{},[49,954,955,956],{},"LM Studio 官网 ",[179,957,958],{"href":958,"rel":959},"https:\u002F\u002Flmstudio.ai\u002F",[183],[49,961,962,963],{},"Codersera — LM Studio Complete Guide 2026 ",[179,964,965],{"href":965,"rel":966},"https:\u002F\u002Fcodersera.com\u002Fblog\u002Flm-studio-complete-guide-2026\u002F",[183],[49,968,969,970],{},"Codersera — Ollama vs LM Studio vs vLLM vs llama.cpp vs MLX 2026 ",[179,971,972],{"href":972,"rel":973},"https:\u002F\u002Fcodersera.com\u002Fblog\u002Follama-vs-lm-studio-vs-vllm-vs-llama-cpp-vs-mlx-2026\u002F",[183],{"title":462,"searchDepth":463,"depth":463,"links":975},[976,977,978,979,980,981,982,983,984,985],{"id":33,"depth":466,"text":34},{"id":44,"depth":466,"text":44},{"id":94,"depth":466,"text":94},{"id":653,"depth":466,"text":654},{"id":165,"depth":466,"text":165},{"id":195,"depth":466,"text":195},{"id":349,"depth":466,"text":349},{"id":384,"depth":466,"text":385},{"id":411,"depth":466,"text":411},{"id":437,"depth":466,"text":437},"\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。",[989,992,995,998],{"q":990,"a":991},"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":993,"a":994},"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":996,"a":997},"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":999,"a":1000},"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。",[497,496],{},[503,504,505],[1005,1008],{"plan":638,"price":509,"features":1006,"notes":1007},"全功能 GUI + Headless API + GGUF\u002FMLX","供个人 \u002F 评估使用",{"plan":644,"price":1009,"features":1010,"notes":1011},"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",[1014,1015],{"name":520,"url":433},{"name":1016,"url":947},"Claude Code 上手",{"power":525,"ux":477,"price":477,"cn_support":463,"stability":525},{"title":216,"description":987},"coding\u002Flocal\u002Flm-studio",[1021,1023,1025],{"name":1022,"url":958,"accessed":531},"LM Studio 官网",{"name":1024,"url":965,"accessed":531},"Codersera — LM Studio Complete Guide 2026",{"name":1026,"url":972,"accessed":531},"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",[476,1030,1031,1032,1033,1034,1035],"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":1040,"title":213,"alternatives":1041,"api_compatible":3,"body":1046,"category":476,"chinese_friendly":477,"cover":1482,"description":1483,"domestic":480,"extension":481,"faq":1484,"free":480,"github":3,"languages":1497,"meta":1498,"models":3,"navigation":499,"notSuitable":3,"opensource":499,"path":17,"pillar":501,"platforms":1499,"priceTable":1502,"pricing":1510,"published":517,"relatedPlaybooks":1511,"relatedReviews":3,"score":1514,"self_host":499,"seo":1515,"slug":1516,"sources":1517,"stem":1524,"suitable":3,"tagline":1525,"tags":1526,"updated":531,"verdict":1529,"website":1530,"__hash__":1531},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Flobe-chat.md",[1042,1043,1044,1045],{"name":554,"url":500},{"name":25,"url":26},{"name":22,"url":23},{"name":19,"url":20},{"type":28,"value":1047,"toc":1470},[1048,1050,1057,1060,1062,1124,1126,1139,1142,1146,1150,1170,1174,1191,1193,1219,1221,1357,1359,1397,1399,1425,1427,1445,1447],[31,1049,34],{"id":33},[36,1051,1052,1053,1056],{},"LobeChat 是 LobeHub 团队的开源 AI 聊天框架，2023 年发布、GitHub 72k+ stars、MIT 协议。",[52,1054,1055],{},"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。",[36,1058,1059],{},"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）。",[31,1061,44],{"id":44},[46,1063,1064,1069,1075,1080,1086,1092,1097,1102,1108,1114],{},[49,1065,1066,1068],{},[52,1067,54],{},"：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",[49,1070,1071,1074],{},[52,1072,1073],{},"多模型对比","：同 prompt 给多模型并排回答",[49,1076,1077,1079],{},[52,1078,60],{},"：上传 PDF \u002F Word \u002F 网页 → 向量化 → 检索引用",[49,1081,1082,1085],{},[52,1083,1084],{},"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",[49,1087,1088,1091],{},[52,1089,1090],{},"助手市场","：几百个预设 AI 角色，一键导入",[49,1093,1094,1096],{},[52,1095,72],{},"：扩展任意工具能力",[49,1098,1099],{},[52,1100,1101],{},"代码解释器 \u002F 文件上传 \u002F TTS \u002F 多模态",[49,1103,1104,1107],{},[52,1105,1106],{},"Web + 桌面 + Docker","：三形态，数据可完全本地",[49,1109,1110,1113],{},[52,1111,1112],{},"LobeHub Cloud","：官方云托管，免部署",[49,1115,1116,1119,1120,1123],{},[52,1117,1118],{},"快捷指令 \u002F 工作流","：自定义 prompt 模板，",[564,1121,1122],{},"\u002Fpodcast-summary"," 类用法",[31,1125,94],{"id":94},[46,1127,1128,1134],{},[49,1129,1130,1133],{},[52,1131,1132],{},"自托管 \u002F 桌面","：完全免费、MIT 开源",[49,1135,1136,1138],{},[52,1137,1112],{},"：订阅制，云端托管 + 团队协作 + 同步",[36,1140,1141],{},"模型 API 费用按你自己的供应商付费；本地 Ollama \u002F LM Studio 零成本。",[31,1143,1145],{"id":1144},"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）",[36,1147,1148],{},[52,1149,123],{},[46,1151,1152,1155,1158,1161,1164,1167],{},[49,1153,1154],{},"界面颜值是这一类工具里第一档（深色 \u002F 透明 \u002F 现代感）",[49,1156,1157],{},"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",[49,1159,1160],{},"知识库 RAG 上传 50+ PDF 后检索准确，引用片段可视化",[49,1162,1163],{},"助手市场拿来即用——「Code Reviewer」「Translation Polish」节省 prompt 编写",[49,1165,1166],{},"Docker 一键部署，团队 5 人共享流畅",[49,1168,1169],{},"多平台数据同步（Cloud \u002F WebDAV）",[36,1171,1172],{},[52,1173,145],{},[46,1175,1176,1179,1182,1185,1188],{},[49,1177,1178],{},"自托管要熟悉 Docker + 反代 + HTTPS",[49,1180,1181],{},"国内连 OpenAI \u002F Claude 需自带网络方案",[49,1183,1184],{},"Web 版数据存 LobeHub，隐私敏感场景走桌面 \u002F Docker",[49,1186,1187],{},"插件市场质量参差，要自筛",[49,1189,1190],{},"团队多人共享需配 LobeHub Cloud 或自建数据库（Postgres + S3）",[31,1192,165],{"id":165},[167,1194,1195,1198,1204,1207,1210,1213,1216],{},[49,1196,1197],{},"选形态：Web（chat.lobehub.com 注册即用） \u002F 桌面（GitHub Releases 下载） \u002F Docker",[49,1199,1200,1201],{},"Docker：",[564,1202,1203],{},"docker run -d -p 3210:3210 -e OPENAI_API_KEY=sk-xxx --name lobe-chat lobehub\u002Flobe-chat",[49,1205,1206],{},"设置 → AI 服务商 → 添加 OpenAI \u002F Claude \u002F DeepSeek \u002F Ollama",[49,1208,1209],{},"模型选择器测试对话",[49,1211,1212],{},"知识库：拖文件 → 等向量化 → 对话引用",[49,1214,1215],{},"助手市场拉「Code Reviewer」「论文翻译润色」试用",[49,1217,1218],{},"进阶：插件市场启用联网搜索 \u002F 代码执行；MCP 自定义工具",[31,1220,195],{"id":195},[197,1222,1223,1237],{},[200,1224,1225],{},[203,1226,1227,1229,1231,1233,1235],{},[206,1228,208],{},[206,1230,213],{},[206,1232,13],{},[206,1234,219],{},[206,1236,216],{},[221,1238,1239,1251,1264,1277,1290,1306,1319,1333,1345],{},[203,1240,1241,1243,1245,1247,1249],{},[226,1242,228],{},[226,1244,1106],{},[226,1246,231],{},[226,1248,239],{},[226,1250,231],{},[203,1252,1253,1255,1258,1260,1262],{},[226,1254,54],{},[226,1256,1257],{},"✅ 80+",[226,1259,269],{},[226,1261,269],{},[226,1263,251],{},[203,1265,1266,1268,1271,1273,1275],{},[226,1267,1073],{},[226,1269,1270],{},"✅ 一等",[226,1272,269],{},[226,1274,266],{},[226,1276,266],{},[203,1278,1279,1281,1283,1285,1288],{},[226,1280,258],{},[226,1282,269],{},[226,1284,269],{},[226,1286,1287],{},"✅ + oikb",[226,1289,266],{},[203,1291,1292,1295,1298,1301,1304],{},[226,1293,1294],{},"插件 \u002F 助手市场",[226,1296,1297],{},"✅ 丰富",[226,1299,1300],{},"300+ 助手",[226,1302,1303],{},"Tools",[226,1305,266],{},[203,1307,1308,1310,1312,1314,1317],{},[226,1309,274],{},[226,1311,269],{},[226,1313,269],{},[226,1315,1316],{},"✅ mcpo",[226,1318,266],{},[203,1320,1321,1324,1327,1329,1331],{},[226,1322,1323],{},"多用户",[226,1325,1326],{},"配 Cloud \u002F 自建",[226,1328,296],{},[226,1330,1270],{},[226,1332,296],{},[203,1334,1335,1337,1339,1341,1343],{},[226,1336,334],{},[226,1338,340],{},[226,1340,337],{},[226,1342,346],{},[226,1344,343],{},[203,1346,1347,1349,1351,1353,1355],{},[226,1348,318],{},[226,1350,324],{},[226,1352,321],{},[226,1354,324],{},[226,1356,327],{},[31,1358,349],{"id":349},[46,1360,1361,1367,1373,1379,1385,1391],{},[49,1362,1363,1366],{},[52,1364,1365],{},"Web 版数据不本地","：隐私敏感选桌面或 Docker 自托管",[49,1368,1369,1372],{},[52,1370,1371],{},"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",[49,1374,1375,1378],{},[52,1376,1377],{},"Docker 自托管暴露公网","：上反代 + HTTPS + Auth + 备份数据库",[49,1380,1381,1384],{},[52,1382,1383],{},"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",[49,1386,1387,1390],{},[52,1388,1389],{},"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",[49,1392,1393,1396],{},[52,1394,1395],{},"同步选 Cloud vs WebDAV","：团队多端走 LobeHub Cloud；个人多设备 WebDAV 即可",[31,1398,385],{"id":384},[46,1400,1401,1404,1407,1410,1413,1416,1419,1422],{},[49,1402,1403],{},"✅ Web + 桌面双形态需求",[49,1405,1406],{},"✅ Docker 自托管 \u002F 团队共享",[49,1408,1409],{},"✅ 多模型对比 \u002F 选型",[49,1411,1412],{},"✅ 中文重度用户",[49,1414,1415],{},"✅ 助手市场 \u002F 插件生态用户",[49,1417,1418],{},"❌ 强企业 RBAC + 多租户（Open WebUI 更完善）",[49,1420,1421],{},"❌ 只要桌面 + 数据完全本地（Cherry Studio 同样优秀）",[49,1423,1424],{},"❌ 完全不会碰 Docker",[31,1426,411],{"id":411},[46,1428,1429,1433,1437,1441],{},[49,1430,1431],{},[179,1432,942],{"href":500},[49,1434,1435],{},[179,1436,937],{"href":26},[49,1438,1439],{},[179,1440,428],{"href":23},[49,1442,1443],{},[179,1444,434],{"href":433},[31,1446,437],{"id":437},[167,1448,1449,1456,1463],{},[49,1450,1451,1452],{},"LobeChat GitHub 仓库（72k+ stars，MIT）",[179,1453,1454],{"href":1454,"rel":1455},"https:\u002F\u002Fgithub.com\u002Flobehub\u002Flobe-chat",[183],[49,1457,1458,1459],{},"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）",[179,1460,1461],{"href":1461,"rel":1462},"https:\u002F\u002Fcloud.tencent.com\u002Fdeveloper\u002Farticle\u002F2622150",[183],[49,1464,1465,1466],{},"Ofox.ai — LobeChat 完全配置指南 2026（2026-04-17）",[179,1467,1468],{"href":1468,"rel":1469},"https:\u002F\u002Fofox.ai\u002Fzh\u002Fblog\u002Flobechat-api-configuration-guide-2026",[183],{"title":462,"searchDepth":463,"depth":463,"links":1471},[1472,1473,1474,1475,1476,1477,1478,1479,1480,1481],{"id":33,"depth":466,"text":34},{"id":44,"depth":466,"text":44},{"id":94,"depth":466,"text":94},{"id":1144,"depth":466,"text":1145},{"id":165,"depth":466,"text":165},{"id":195,"depth":466,"text":195},{"id":349,"depth":466,"text":349},{"id":384,"depth":466,"text":385},{"id":411,"depth":466,"text":411},{"id":437,"depth":466,"text":437},"\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。",[1485,1488,1491,1494],{"q":1486,"a":1487},"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":1489,"a":1490},"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":1492,"a":1493},"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":1495,"a":1496},"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。",[496,497],{},[1500,503,504,505,1501],"web","docker",[1503,1506],{"plan":1132,"price":509,"features":1504,"notes":1505},"全功能 \u002F 80+ 模型 \u002F 知识库 \u002F 插件 \u002F 助手市场","MIT 协议",{"plan":1112,"price":1507,"features":1508,"notes":1509},"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",[1512,1513],{"name":520,"url":433},{"name":1016,"url":947},{"power":477,"ux":477,"price":477,"cn_support":477,"stability":525},{"title":213,"description":1483},"coding\u002Flocal\u002Flobe-chat",[1518,1520,1522],{"name":1519,"url":1454,"accessed":531},"LobeChat GitHub",{"name":1521,"url":1461,"accessed":531},"腾讯云开发者社区 — Lobe Chat 终极桌面客户端",{"name":1523,"url":1468,"accessed":531},"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",[476,1500,539,540,542,1527,1528,543],"plugin","mcp","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",{"id":1533,"title":739,"alternatives":1534,"api_compatible":3,"body":1539,"category":476,"chinese_friendly":463,"cover":1958,"description":1959,"domestic":480,"extension":481,"faq":1960,"free":480,"github":3,"languages":1973,"meta":1974,"models":3,"navigation":499,"notSuitable":3,"opensource":499,"path":23,"pillar":501,"platforms":1975,"priceTable":1976,"pricing":1980,"published":517,"relatedPlaybooks":1981,"relatedReviews":3,"score":1984,"self_host":499,"seo":1985,"slug":1986,"sources":1987,"stem":1993,"suitable":3,"tagline":1994,"tags":1995,"updated":531,"verdict":2000,"website":2001,"__hash__":2002},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Follama.md",[1535,1536,1537,1538],{"name":19,"url":20},{"name":25,"url":26},{"name":554,"url":500},{"name":16,"url":17},{"type":28,"value":1540,"toc":1946},[1541,1543,1550,1553,1555,1623,1625,1628,1632,1636,1656,1660,1691,1693,1727,1729,1844,1846,1878,1880,1903,1905,1923,1925],[31,1542,34],{"id":33},[36,1544,1545,1546,1549],{},"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，",[564,1547,1548],{},"ollama pull"," 一键拉。",[36,1551,1552],{},"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）。",[31,1554,44],{"id":44},[46,1556,1557,1563,1571,1577,1584,1599,1605,1611,1617],{},[49,1558,1559,1562],{},[52,1560,1561],{},"后台 Daemon","：开机自启，应用调用零延迟",[49,1564,1565,598,1568],{},[52,1566,1567],{},"CLI",[564,1569,1570],{},"ollama pull \u002F run \u002F list \u002F show \u002F create \u002F serve",[49,1572,1573,1576],{},[52,1574,1575],{},"Modelfile","：类 Dockerfile 注册任意 GGUF，配 SYSTEM \u002F PARAMETER \u002F TEMPLATE",[49,1578,1579,598,1581],{},[52,1580,798],{},[564,1582,1583],{},"http:\u002F\u002Flocalhost:11434\u002Fv1\u002Fchat\u002Fcompletions",[49,1585,1586,598,1589,1592,1593,1592,1596],{},[52,1587,1588],{},"原生 API",[564,1590,1591],{},"\u002Fapi\u002Fchat","、",[564,1594,1595],{},"\u002Fapi\u002Fgenerate",[564,1597,1598],{},"\u002Fapi\u002Fembeddings",[49,1600,1601,1604],{},[52,1602,1603],{},"模型库","：官方注册表内置 Llama \u002F Qwen \u002F DeepSeek \u002F Gemma \u002F Mistral \u002F GPT-OSS 等",[49,1606,1607,1610],{},[52,1608,1609],{},"MLX 加速（Mac）","：0.19+ 起 M 系列自动用 MLX",[49,1612,1613,1616],{},[52,1614,1615],{},"量化","：默认 Q4_K_M、支持 Q5 \u002F Q8 \u002F FP16",[49,1618,1619,1622],{},[52,1620,1621],{},"跨平台","：Win \u002F Mac \u002F Linux 安装包，Docker 官方镜像",[31,1624,94],{"id":94},[36,1626,1627],{},"完全免费、MIT 开源、商用免费。",[31,1629,1631],{"id":1630},"实测m2-pro-qwen3-coder-7b-q4","实测（M2 Pro + Qwen3-Coder-7B Q4）",[36,1633,1634],{},[52,1635,123],{},[46,1637,1638,1644,1647,1650,1653],{},[49,1639,1640,1643],{},[564,1641,1642],{},"ollama run qwen3-coder:7b"," 一行起飞，3 秒进交互",[49,1645,1646],{},"REST API 配 Cursor \u002F Cline \u002F Continue 几乎全工具开箱即用",[49,1648,1649],{},"Modelfile 写自定义编码助手（low temperature + system prompt + 16K context）几分钟搞定",[49,1651,1652],{},"多模型并存，按需切换，内存占用合理",[49,1654,1655],{},"Mac M 系列 MLX 后比旧 GGUF 模式快显著",[36,1657,1658],{},[52,1659,145],{},[46,1661,1662,1672,1678,1685,1688],{},[49,1663,1664,1665,1668,1669],{},"默认 ",[564,1666,1667],{},"num_ctx"," 偏小（2048），跑长上下文要在 Modelfile 加 ",[564,1670,1671],{},"PARAMETER num_ctx 16384",[49,1673,1674,1675],{},"模型默认走 0.0.0.0:11434 ↔ Docker 容器互访要 ",[564,1676,1677],{},"--add-host=host.docker.internal:host-gateway",[49,1679,1680,1681,1684],{},"国内 ",[564,1682,1683],{},"ollama.com\u002Flibrary"," 下载偶有慢，可手动 HF 下 GGUF + Modelfile 自建",[49,1686,1687],{},"多用户并发吞吐显著低于 vLLM",[49,1689,1690],{},"没有 GUI，模型浏览 \u002F 参数面板要走 LM Studio \u002F Open WebUI 配合",[31,1692,165],{"id":165},[167,1694,1695,1701,1707,1712,1718,1724],{},[49,1696,1697,1700],{},[564,1698,1699],{},"curl -fsSL https:\u002F\u002Follama.ai\u002Finstall.sh | sh","（Mac \u002F Linux）；Windows winget",[49,1702,1703,1706],{},[564,1704,1705],{},"ollama pull qwen3-coder:7b","（按需换模型）",[49,1708,1709,1711],{},[564,1710,1642],{}," 直接聊",[49,1713,1714,1715],{},"应用接入：baseURL = ",[564,1716,1717],{},"http:\u002F\u002Flocalhost:11434\u002Fv1",[49,1719,1720,1721],{},"自定义：写 Modelfile → ",[564,1722,1723],{},"ollama create my-coder -f Modelfile",[49,1725,1726],{},"进阶：装 Open WebUI 做前端 \u002F 多人共享",[31,1728,195],{"id":195},[197,1730,1731,1746],{},[200,1732,1733],{},[203,1734,1735,1737,1739,1741,1744],{},[206,1736,208],{},[206,1738,739],{},[206,1740,216],{},[206,1742,1743],{},"vLLM",[206,1745,744],{},[221,1747,1748,1764,1776,1789,1802,1818,1830],{},[203,1749,1750,1752,1755,1758,1761],{},[226,1751,228],{},[226,1753,1754],{},"CLI + Daemon",[226,1756,1757],{},"GUI + Headless",[226,1759,1760],{},"Python Server",[226,1762,1763],{},"C++ 二进制",[203,1765,1766,1768,1770,1772,1774],{},[226,1767,165],{},[226,1769,857],{},[226,1771,857],{},[226,1773,836],{},[226,1775,865],{},[203,1777,1778,1780,1782,1785,1787],{},[226,1779,767],{},[226,1781,1567],{},[226,1783,1784],{},"✅ GUI",[226,1786,296],{},[226,1788,296],{},[203,1790,1791,1794,1796,1798,1800],{},[226,1792,1793],{},"OpenAI 兼容",[226,1795,804],{},[226,1797,801],{},[226,1799,269],{},[226,1801,269],{},[203,1803,1804,1807,1810,1813,1816],{},[226,1805,1806],{},"多用户吞吐",[226,1808,1809],{},"弱（~40 tok\u002Fs）",[226,1811,1812],{},"中（50–90）",[226,1814,1815],{},"强（800–12500）",[226,1817,836],{},[203,1819,1820,1822,1824,1826,1828],{},[226,1821,813],{},[226,1823,818],{},[226,1825,269],{},[226,1827,791],{},[226,1829,343],{},[203,1831,1832,1834,1836,1839,1842],{},[226,1833,841],{},[226,1835,324],{},[226,1837,1838],{},"闭源",[226,1840,1841],{},"Apache 2.0",[226,1843,324],{},[31,1845,349],{"id":349},[46,1847,1848,1854,1860,1866,1872],{},[49,1849,1850,1853],{},[52,1851,1852],{},"num_ctx 一定要设","：默认 2K 太小，跑代码 \u002F 长文档要 16K+",[49,1855,1856,1859],{},[52,1857,1858],{},"Modelfile 模板别漏 TEMPLATE","：错的 chat template 会让模型输出乱码 \u002F 不停",[49,1861,1862,1865],{},[52,1863,1864],{},"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×",[49,1867,1868,1871],{},[52,1869,1870],{},"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",[49,1873,1874,1877],{},[52,1875,1876],{},"Mac 让它自动用 MLX","：升 0.19+；不要手动强制 GGUF + Metal",[31,1879,385],{"id":384},[46,1881,1882,1885,1888,1891,1894,1897,1900],{},[49,1883,1884],{},"✅ 应用 \u002F IDE 接本地模型（Cursor \u002F Cline \u002F Continue）",[49,1886,1887],{},"✅ 个人 \u002F 评估 \u002F 脚本自动化",[49,1889,1890],{},"✅ Modelfile 自定义系统 prompt + 参数",[49,1892,1893],{},"✅ Mac M 系列 MLX 用户",[49,1895,1896],{},"❌ 多用户并发生产服务（用 vLLM）",[49,1898,1899],{},"❌ GUI 调参 \u002F 模型浏览（配 LM Studio \u002F Open WebUI）",[49,1901,1902],{},"❌ 极致单卡吞吐研究（直接 llama.cpp \u002F vLLM）",[31,1904,411],{"id":411},[46,1906,1907,1911,1915,1919],{},[49,1908,1909],{},[179,1910,423],{"href":20},[49,1912,1913],{},[179,1914,937],{"href":26},[49,1916,1917],{},[179,1918,942],{"href":500},[49,1920,1921],{},[179,1922,434],{"href":433},[31,1924,437],{"id":437},[167,1926,1927,1934,1941],{},[49,1928,1929,1930],{},"Markaicode — Import GGUF Models into Ollama 2026（2026-05-15）",[179,1931,1932],{"href":1932,"rel":1933},"https:\u002F\u002Fmarkaicode.com\u002Fimport-gguf-models-ollama-guide",[183],[49,1935,1936,1937],{},"ComputingForGeeks — Ollama Models Cheat Sheet 2026 ",[179,1938,1939],{"href":1939,"rel":1940},"https:\u002F\u002Fcomputingforgeeks.com\u002Follama-models-cheat-sheet",[183],[49,1942,969,1943],{},[179,1944,972],{"href":972,"rel":1945},[183],{"title":462,"searchDepth":463,"depth":463,"links":1947},[1948,1949,1950,1951,1952,1953,1954,1955,1956,1957],{"id":33,"depth":466,"text":34},{"id":44,"depth":466,"text":44},{"id":94,"depth":466,"text":94},{"id":1630,"depth":466,"text":1631},{"id":165,"depth":466,"text":165},{"id":195,"depth":466,"text":195},{"id":349,"depth":466,"text":349},{"id":384,"depth":466,"text":385},{"id":411,"depth":466,"text":411},{"id":437,"depth":466,"text":437},"\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。",[1961,1964,1967,1970],{"q":1962,"a":1963},"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":1965,"a":1966},"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":1968,"a":1969},"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":1971,"a":1972},"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。",[497],{},[503,504,505,1501],[1977],{"plan":101,"price":509,"features":1978,"notes":1979},"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）",[1982,1983],{"name":520,"url":433},{"name":1016,"url":947},{"power":525,"ux":525,"price":477,"cn_support":463,"stability":477},{"title":739,"description":1959},"coding\u002Flocal\u002Follama",[1988,1990,1992],{"name":1989,"url":1932,"accessed":531},"Markaicode — Import GGUF 2026",{"name":1991,"url":1939,"accessed":531},"ComputingForGeeks — Ollama Cheat Sheet 2026",{"name":1026,"url":972,"accessed":531},"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",[476,1996,1997,1998,1999,1031,1032,1035,543],"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":2004,"title":219,"alternatives":2005,"api_compatible":3,"body":2010,"category":476,"chinese_friendly":525,"cover":2425,"description":2426,"domestic":480,"extension":481,"faq":2427,"free":480,"github":3,"languages":2440,"meta":2441,"models":3,"navigation":499,"notSuitable":3,"opensource":499,"path":26,"pillar":501,"platforms":2442,"priceTable":2444,"pricing":2451,"published":517,"relatedPlaybooks":2452,"relatedReviews":3,"score":2455,"self_host":499,"seo":2456,"slug":2457,"sources":2458,"stem":2465,"suitable":3,"tagline":2466,"tags":2467,"updated":531,"verdict":2470,"website":2471,"__hash__":2472},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Fopen-webui.md",[2006,2007,2008,2009],{"name":16,"url":17},{"name":554,"url":500},{"name":22,"url":23},{"name":19,"url":20},{"type":28,"value":2011,"toc":2413},[2012,2014,2017,2020,2022,2084,2086,2089,2093,2097,2125,2129,2153,2155,2189,2191,2298,2300,2343,2345,2368,2370,2388,2390],[31,2013,34],{"id":33},[36,2015,2016],{},"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。",[36,2018,2019],{},"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。",[31,2021,44],{"id":44},[46,2023,2024,2030,2036,2042,2048,2054,2060,2066,2072,2078],{},[49,2025,2026,2029],{},[52,2027,2028],{},"多模型后端","：Ollama \u002F OpenAI \u002F vLLM \u002F Anthropic \u002F Groq \u002F LocalAI \u002F 任意 OpenAI 兼容",[49,2031,2032,2035],{},[52,2033,2034],{},"多用户 + RBAC","：注册 \u002F 邀请 \u002F 角色权限 \u002F 工作区隔离",[49,2037,2038,2041],{},[52,2039,2040],{},"RAG 知识库","：上传文档 \u002F 网址 \u002F SearXNG 联网搜索 → 向量化 → 对话引用",[49,2043,2044,2047],{},[52,2045,2046],{},"Tools \u002F Functions","：Python 写函数即扩展（联网 \u002F 计算器 \u002F 自定义 API）",[49,2049,2050,2053],{},[52,2051,2052],{},"mcpo","：MCP-to-OpenAPI 代理，任意 MCP 服务器接进来",[49,2055,2056,2059],{},[52,2057,2058],{},"oikb","：知识库同步本地文件夹 \u002F GitHub \u002F S3 \u002F Confluence 等 40+ 源",[49,2061,2062,2065],{},[52,2063,2064],{},"open-terminal \u002F cptr","：给 AI 真实终端 + 文件 + 沙箱执行",[49,2067,2068,2071],{},[52,2069,2070],{},"图像生成","：Stable Diffusion \u002F DALL·E \u002F 自托管接入",[49,2073,2074,2077],{},[52,2075,2076],{},"语音输入 \u002F TTS","：内置",[49,2079,2080,2083],{},[52,2081,2082],{},"企业 LTS","：custom branding + SLA + 长期支持版本（联系销售）",[31,2085,94],{"id":94},[36,2087,2088],{},"完全免费、MIT 开源、商用免费。Enterprise 提供品牌定制 + SLA + LTS。",[31,2090,2092],{"id":2091},"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）",[36,2094,2095],{},[52,2096,123],{},[46,2098,2099,2106,2109,2116,2119,2122],{},[49,2100,2101,2102,2105],{},"单条 ",[564,2103,2104],{},"docker run"," 五分钟上线",[49,2107,2108],{},"自带的多用户 + 角色权限省去重新搭 Auth",[49,2110,2111,2112,2115],{},"RAG 直传 30 个 PDF 后向量化顺利，对话中 ",[564,2113,2114],{},"#知识库"," 引用准确",[49,2117,2118],{},"mcpo 把 GitHub MCP 服务器接进来，团队对话里直接 issue \u002F PR 操作",[49,2120,2121],{},"模型切换流畅，OpenAI + Ollama 并存",[49,2123,2124],{},"SearXNG 联网搜索给模型实时信息，过时知识截止问题缓解",[36,2126,2127],{},[52,2128,145],{},[46,2130,2131,2134,2140,2147,2150],{},[49,2132,2133],{},"Docker 镜像 ~1.5GB，首次拉取偏慢",[49,2135,1664,2136,2139],{},[564,2137,2138],{},"0.0.0.0"," 公网暴露要加 HTTPS + 反代",[49,2141,2142,2143,2146],{},"嵌入模型 ",[564,2144,2145],{},"sentence-transformers"," 中文效果一般，建议换 bge-m3",[49,2148,2149],{},"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）",[49,2151,2152],{},"版本升级要看 changelog，部分 minor 含 breaking 改动",[31,2154,165],{"id":165},[167,2156,2157,2163,2170,2173,2176,2179,2182],{},[49,2158,2159,2160],{},"装 Docker → ",[564,2161,2162],{},"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",[49,2164,2165,2166,2169],{},"浏览器开 ",[564,2167,2168],{},"http:\u002F\u002Flocalhost:3000"," → 注册第一个账号（管理员）",[49,2171,2172],{},"设置 → Connections → 连接 Ollama \u002F 加 OpenAI Key",[49,2174,2175],{},"Models → Pull \u002F Discover 模型",[49,2177,2178],{},"Workspaces → 建知识库 → 上传文档",[49,2180,2181],{},"Tools → 启用 \u002F 写自定义函数",[49,2183,2184,2185,2188],{},"生产部署：Nginx 反代 + Let's Encrypt + 备份 ",[564,2186,2187],{},"\u002Fapp\u002Fbackend\u002Fdata"," volume",[31,2190,195],{"id":195},[197,2192,2193,2207],{},[200,2194,2195],{},[203,2196,2197,2199,2201,2203,2205],{},[206,2198,208],{},[206,2200,219],{},[206,2202,213],{},[206,2204,13],{},[206,2206,216],{},[221,2208,2209,2221,2233,2247,2260,2274,2286],{},[203,2210,2211,2213,2215,2217,2219],{},[226,2212,228],{},[226,2214,239],{},[226,2216,234],{},[226,2218,231],{},[226,2220,231],{},[203,2222,2223,2225,2227,2229,2231],{},[226,2224,1323],{},[226,2226,1270],{},[226,2228,269],{},[226,2230,296],{},[226,2232,296],{},[203,2234,2235,2238,2241,2243,2245],{},[226,2236,2237],{},"RAG",[226,2239,2240],{},"✅ 强 + oikb",[226,2242,269],{},[226,2244,269],{},[226,2246,266],{},[203,2248,2249,2252,2254,2256,2258],{},[226,2250,2251],{},"工具 \u002F MCP",[226,2253,1316],{},[226,2255,269],{},[226,2257,269],{},[226,2259,266],{},[203,2261,2262,2265,2268,2270,2272],{},[226,2263,2264],{},"自托管",[226,2266,2267],{},"✅ Docker \u002F K8s",[226,2269,293],{},[226,2271,296],{},[226,2273,296],{},[203,2275,2276,2278,2280,2282,2284],{},[226,2277,334],{},[226,2279,346],{},[226,2281,340],{},[226,2283,337],{},[226,2285,343],{},[203,2287,2288,2290,2292,2294,2296],{},[226,2289,318],{},[226,2291,324],{},[226,2293,324],{},[226,2295,321],{},[226,2297,327],{},[31,2299,349],{"id":349},[46,2301,2302,2308,2316,2325,2331,2337],{},[49,2303,2304,2307],{},[52,2305,2306],{},"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",[49,2309,2310,2315],{},[52,2311,2312,2313,2188],{},"备份 ",[564,2314,2187],{},"：知识库 \u002F 用户 \u002F 对话全在里面",[49,2317,2318,2321,2322,2324],{},[52,2319,2320],{},"中文 RAG 换嵌入模型","：默认 ",[564,2323,2145],{}," 中文一般，配 bge-m3 或硅基流动嵌入 API",[49,2326,2327,2330],{},[52,2328,2329],{},"mcpo 工具范围谨慎","：MCP 给 AI 真实能力，第三方服务器审一遍",[49,2332,2333,2336],{},[52,2334,2335],{},"后端吞吐看 Ollama","：5+ 并发上 vLLM 后端，Ollama 单 worker 会排队",[49,2338,2339,2342],{},[52,2340,2341],{},"升级前看 changelog","：weekly 更新，偶有 breaking",[31,2344,385],{"id":384},[46,2346,2347,2350,2353,2356,2359,2362,2365],{},[49,2348,2349],{},"✅ 团队 \u002F 家庭 \u002F 公司多人共享 AI 平台",[49,2351,2352],{},"✅ 要 Web 端访问 \u002F 移动端兼容",[49,2354,2355],{},"✅ 自托管 \u002F 完全控制数据",[49,2357,2358],{},"✅ MCP \u002F 工具调用刚需",[49,2360,2361],{},"❌ 单人桌面体验（用 Cherry Studio）",[49,2363,2364],{},"❌ 零运维 \u002F 不愿碰 Docker",[49,2366,2367],{},"❌ iOS 原生 App 主力",[31,2369,411],{"id":411},[46,2371,2372,2376,2380,2384],{},[49,2373,2374],{},[179,2375,418],{"href":17},[49,2377,2378],{},[179,2379,942],{"href":500},[49,2381,2382],{},[179,2383,428],{"href":23},[49,2385,2386],{},[179,2387,434],{"href":433},[31,2389,437],{"id":437},[167,2391,2392,2399,2406],{},[49,2393,2394,2395],{},"Open WebUI 官方文档 ",[179,2396,2397],{"href":2397,"rel":2398},"https:\u002F\u002Fdocs.openwebui.com\u002F",[183],[49,2400,2401,2402],{},"Local AI Master — Open WebUI Setup Guide 2026 ",[179,2403,2404],{"href":2404,"rel":2405},"https:\u002F\u002Flocalaimaster.com\u002Fblog\u002Fopen-webui-setup-guide",[183],[49,2407,2408,2409],{},"AIToolDiscovery — Set Up Open-WebUI with Ollama 2026 ",[179,2410,2411],{"href":2411,"rel":2412},"https:\u002F\u002Fwww.aitooldiscovery.com\u002Fhow-to\u002Fsetup-open-webui-ollama",[183],{"title":462,"searchDepth":463,"depth":463,"links":2414},[2415,2416,2417,2418,2419,2420,2421,2422,2423,2424],{"id":33,"depth":466,"text":34},{"id":44,"depth":466,"text":44},{"id":94,"depth":466,"text":94},{"id":2091,"depth":466,"text":2092},{"id":165,"depth":466,"text":165},{"id":195,"depth":466,"text":195},{"id":349,"depth":466,"text":349},{"id":384,"depth":466,"text":385},{"id":411,"depth":466,"text":411},{"id":437,"depth":466,"text":437},"\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。",[2428,2431,2434,2437],{"q":2429,"a":2430},"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":2432,"a":2433},"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":2435,"a":2436},"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":2438,"a":2439},"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。",[497,496],{},[1501,505,504,503,2443],"kubernetes",[2445,2447],{"plan":101,"price":509,"features":2446,"notes":1505},"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":107,"price":2448,"features":2449,"notes":2450},"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",[2453,2454],{"name":520,"url":433},{"name":1016,"url":947},{"power":477,"ux":525,"price":477,"cn_support":525,"stability":477},{"title":219,"description":2426},"coding\u002Flocal\u002Fopen-webui",[2459,2461,2463],{"name":2460,"url":2397,"accessed":531},"Open WebUI 官方文档",{"name":2462,"url":2404,"accessed":531},"Local AI Master — Open WebUI Setup Guide 2026",{"name":2464,"url":2411,"accessed":531},"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",[476,2468,1501,542,2469,22,543],"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",1782316489324]