[{"data":1,"prerenderedAt":2834},["ShallowReactive",2],{"header-counts":3,"footer-counts":6,"tool-tools\u002Fcoding\u002Flocal\u002Follama":9,"tool-stats-coding\u002Flocal\u002Follama":574,"cat-rank-coding-local":577,"tool-related-coding\u002Flocal\u002Follama":2833},{"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":504,"chinese_friendly":491,"cover":505,"description":506,"domestic":507,"extension":508,"faq":509,"free":507,"github":25,"languages":522,"meta":524,"models":25,"navigation":525,"notSuitable":25,"opensource":525,"path":526,"pillar":527,"platforms":528,"priceTable":533,"pricing":539,"published":540,"relatedPlaybooks":541,"relatedReviews":25,"score":547,"self_host":525,"seo":550,"slug":551,"sources":552,"stem":560,"suitable":25,"tagline":561,"tags":562,"updated":555,"verdict":571,"website":572,"__hash__":573},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Follama.md","Ollama",[13,16,19,22],{"name":14,"url":15},"lm-studio","\u002Ftools\u002Fcoding\u002Flocal\u002Flm-studio",{"name":17,"url":18},"open-webui","\u002Ftools\u002Fcoding\u002Flocal\u002Fopen-webui",{"name":20,"url":21},"cherry-studio","\u002Ftools\u002Fcoding\u002Flocal\u002Fcherry-studio",{"name":23,"url":24},"lobe-chat","\u002Ftools\u002Fcoding\u002Flocal\u002Flobe-chat",null,{"type":27,"value":28,"toc":489},"minimark",[29,34,43,46,49,122,125,128,132,137,157,162,193,196,231,234,373,376,408,412,435,438,462,465],[30,31,33],"h2",{"id":32},"tldr","TL;DR",[35,36,37,38,42],"p",{},"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，",[39,40,41],"code",{},"ollama pull"," 一键拉。",[35,44,45],{},"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）。",[30,47,48],{"id":48},"核心能力",[50,51,52,60,69,75,83,98,104,110,116],"ul",{},[53,54,55,59],"li",{},[56,57,58],"strong",{},"后台 Daemon","：开机自启，应用调用零延迟",[53,61,62,65,66],{},[56,63,64],{},"CLI","：",[39,67,68],{},"ollama pull \u002F run \u002F list \u002F show \u002F create \u002F serve",[53,70,71,74],{},[56,72,73],{},"Modelfile","：类 Dockerfile 注册任意 GGUF，配 SYSTEM \u002F PARAMETER \u002F TEMPLATE",[53,76,77,65,80],{},[56,78,79],{},"OpenAI 兼容 API",[39,81,82],{},"http:\u002F\u002Flocalhost:11434\u002Fv1\u002Fchat\u002Fcompletions",[53,84,85,65,88,91,92,91,95],{},[56,86,87],{},"原生 API",[39,89,90],{},"\u002Fapi\u002Fchat","、",[39,93,94],{},"\u002Fapi\u002Fgenerate",[39,96,97],{},"\u002Fapi\u002Fembeddings",[53,99,100,103],{},[56,101,102],{},"模型库","：官方注册表内置 Llama \u002F Qwen \u002F DeepSeek \u002F Gemma \u002F Mistral \u002F GPT-OSS 等",[53,105,106,109],{},[56,107,108],{},"MLX 加速（Mac）","：0.19+ 起 M 系列自动用 MLX",[53,111,112,115],{},[56,113,114],{},"量化","：默认 Q4_K_M、支持 Q5 \u002F Q8 \u002F FP16",[53,117,118,121],{},[56,119,120],{},"跨平台","：Win \u002F Mac \u002F Linux 安装包，Docker 官方镜像",[30,123,124],{"id":124},"价格",[35,126,127],{},"完全免费、MIT 开源、商用免费。",[30,129,131],{"id":130},"实测m2-pro-qwen3-coder-7b-q4","实测（M2 Pro + Qwen3-Coder-7B Q4）",[35,133,134],{},[56,135,136],{},"亮点：",[50,138,139,145,148,151,154],{},[53,140,141,144],{},[39,142,143],{},"ollama run qwen3-coder:7b"," 一行起飞，3 秒进交互",[53,146,147],{},"REST API 配 Cursor \u002F Cline \u002F Continue 几乎全工具开箱即用",[53,149,150],{},"Modelfile 写自定义编码助手（low temperature + system prompt + 16K context）几分钟搞定",[53,152,153],{},"多模型并存，按需切换，内存占用合理",[53,155,156],{},"Mac M 系列 MLX 后比旧 GGUF 模式快显著",[35,158,159],{},[56,160,161],{},"踩坑：",[50,163,164,174,180,187,190],{},[53,165,166,167,170,171],{},"默认 ",[39,168,169],{},"num_ctx"," 偏小（2048），跑长上下文要在 Modelfile 加 ",[39,172,173],{},"PARAMETER num_ctx 16384",[53,175,176,177],{},"模型默认走 0.0.0.0:11434 ↔ Docker 容器互访要 ",[39,178,179],{},"--add-host=host.docker.internal:host-gateway",[53,181,182,183,186],{},"国内 ",[39,184,185],{},"ollama.com\u002Flibrary"," 下载偶有慢，可手动 HF 下 GGUF + Modelfile 自建",[53,188,189],{},"多用户并发吞吐显著低于 vLLM",[53,191,192],{},"没有 GUI，模型浏览 \u002F 参数面板要走 LM Studio \u002F Open WebUI 配合",[30,194,195],{"id":195},"上手",[197,198,199,205,211,216,222,228],"ol",{},[53,200,201,204],{},[39,202,203],{},"curl -fsSL https:\u002F\u002Follama.ai\u002Finstall.sh | sh","（Mac \u002F Linux）；Windows winget",[53,206,207,210],{},[39,208,209],{},"ollama pull qwen3-coder:7b","（按需换模型）",[53,212,213,215],{},[39,214,143],{}," 直接聊",[53,217,218,219],{},"应用接入：baseURL = ",[39,220,221],{},"http:\u002F\u002Flocalhost:11434\u002Fv1",[53,223,224,225],{},"自定义：写 Modelfile → ",[39,226,227],{},"ollama create my-coder -f Modelfile",[53,229,230],{},"进阶：装 Open WebUI 做前端 \u002F 多人共享",[30,232,233],{"id":233},"对比",[235,236,237,258],"table",{},[238,239,240],"thead",{},[241,242,243,247,249,252,255],"tr",{},[244,245,246],"th",{},"维度",[244,248,11],{},[244,250,251],{},"LM Studio",[244,253,254],{},"vLLM",[244,256,257],{},"llama.cpp",[259,260,261,279,294,309,325,341,357],"tbody",{},[241,262,263,267,270,273,276],{},[264,265,266],"td",{},"形态",[264,268,269],{},"CLI + Daemon",[264,271,272],{},"GUI + Headless",[264,274,275],{},"Python Server",[264,277,278],{},"C++ 二进制",[241,280,281,283,286,288,291],{},[264,282,195],{},[264,284,285],{},"极低",[264,287,285],{},[264,289,290],{},"中",[264,292,293],{},"高",[241,295,296,299,301,304,307],{},[264,297,298],{},"模型浏览",[264,300,64],{},[264,302,303],{},"✅ GUI",[264,305,306],{},"无",[264,308,306],{},[241,310,311,314,317,320,323],{},[264,312,313],{},"OpenAI 兼容",[264,315,316],{},"✅ :11434",[264,318,319],{},"✅ :1234",[264,321,322],{},"✅",[264,324,322],{},[241,326,327,330,333,336,339],{},[264,328,329],{},"多用户吞吐",[264,331,332],{},"弱（~40 tok\u002Fs）",[264,334,335],{},"中（50–90）",[264,337,338],{},"强（800–12500）",[264,340,290],{},[241,342,343,346,349,351,354],{},[264,344,345],{},"MLX (Mac)",[264,347,348],{},"✅ 0.19+",[264,350,322],{},[264,352,353],{},"部分",[264,355,356],{},"–",[241,358,359,362,365,368,371],{},[264,360,361],{},"开源",[264,363,364],{},"MIT",[264,366,367],{},"闭源",[264,369,370],{},"Apache 2.0",[264,372,364],{},[30,374,375],{"id":375},"避坑",[50,377,378,384,390,396,402],{},[53,379,380,383],{},[56,381,382],{},"num_ctx 一定要设","：默认 2K 太小，跑代码 \u002F 长文档要 16K+",[53,385,386,389],{},[56,387,388],{},"Modelfile 模板别漏 TEMPLATE","：错的 chat template 会让模型输出乱码 \u002F 不停",[53,391,392,395],{},[56,393,394],{},"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×",[53,397,398,401],{},[56,399,400],{},"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",[53,403,404,407],{},[56,405,406],{},"Mac 让它自动用 MLX","：升 0.19+；不要手动强制 GGUF + Metal",[30,409,411],{"id":410},"适合-不适合","适合 \u002F 不适合",[50,413,414,417,420,423,426,429,432],{},[53,415,416],{},"✅ 应用 \u002F IDE 接本地模型（Cursor \u002F Cline \u002F Continue）",[53,418,419],{},"✅ 个人 \u002F 评估 \u002F 脚本自动化",[53,421,422],{},"✅ Modelfile 自定义系统 prompt + 参数",[53,424,425],{},"✅ Mac M 系列 MLX 用户",[53,427,428],{},"❌ 多用户并发生产服务（用 vLLM）",[53,430,431],{},"❌ GUI 调参 \u002F 模型浏览（配 LM Studio \u002F Open WebUI）",[53,433,434],{},"❌ 极致单卡吞吐研究（直接 llama.cpp \u002F vLLM）",[30,436,437],{"id":437},"相关阅读",[50,439,440,446,451,456],{},[53,441,442],{},[443,444,445],"a",{"href":15},"LM Studio 评测",[53,447,448],{},[443,449,450],{"href":18},"Open WebUI 评测",[53,452,453],{},[443,454,455],{"href":21},"Cherry Studio 评测",[53,457,458],{},[443,459,461],{"href":460},"\u002Fplaybook\u002Fonboarding\u002Frag-pipeline-build","RAG Pipeline 搭建 Playbook",[30,463,464],{"id":464},"来源",[197,466,467,475,482],{},[53,468,469,470],{},"Markaicode — Import GGUF Models into Ollama 2026（2026-05-15）",[443,471,472],{"href":472,"rel":473},"https:\u002F\u002Fmarkaicode.com\u002Fimport-gguf-models-ollama-guide",[474],"nofollow",[53,476,477,478],{},"ComputingForGeeks — Ollama Models Cheat Sheet 2026 ",[443,479,480],{"href":480,"rel":481},"https:\u002F\u002Fcomputingforgeeks.com\u002Follama-models-cheat-sheet",[474],[53,483,484,485],{},"Codersera — Ollama vs LM Studio vs vLLM vs llama.cpp vs MLX 2026 ",[443,486,487],{"href":487,"rel":488},"https:\u002F\u002Fcodersera.com\u002Fblog\u002Follama-vs-lm-studio-vs-vllm-vs-llama-cpp-vs-mlx-2026\u002F",[474],{"title":490,"searchDepth":491,"depth":491,"links":492},"",3,[493,495,496,497,498,499,500,501,502,503],{"id":32,"depth":494,"text":33},2,{"id":48,"depth":494,"text":48},{"id":124,"depth":494,"text":124},{"id":130,"depth":494,"text":131},{"id":195,"depth":494,"text":195},{"id":233,"depth":494,"text":233},{"id":375,"depth":494,"text":375},{"id":410,"depth":494,"text":411},{"id":437,"depth":494,"text":437},{"id":464,"depth":494,"text":464},"local","\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。",false,"md",[510,513,516,519],{"q":511,"a":512},"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":514,"a":515},"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":517,"a":518},"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":520,"a":521},"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。",[523],"en",{},true,"\u002Ftools\u002Fcoding\u002Flocal\u002Follama","coding",[529,530,531,532],"windows","macos","linux","docker",[534],{"plan":535,"price":536,"features":537,"notes":538},"开源版","免费","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）","2026-06-19",[542,544],{"name":543,"url":460},"RAG Pipeline 搭建",{"name":545,"url":546},"Claude Code 上手","\u002Fplaybook\u002Fonboarding\u002Fclaude-code-getting-started",{"power":548,"ux":548,"price":549,"cn_support":491,"stability":549},4,5,{"title":11,"description":506},"coding\u002Flocal\u002Follama",[553,556,558],{"name":554,"url":472,"accessed":555},"Markaicode — Import GGUF 2026","2026-06-24",{"name":557,"url":480,"accessed":555},"ComputingForGeeks — Ollama Cheat Sheet 2026",{"name":559,"url":487,"accessed":555},"Codersera — Ollama vs LM Studio vs vLLM 2026","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",[504,563,564,565,566,567,568,569,570],"daemon","cli","rest-api","modelfile","gguf","mlx","openai-compatible","open-source","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",{"ok":525,"slug":575,"viewCount":576,"clickCount":576,"avgRating":576,"ratingCount":576},"coding%2Flocal%2Follama",0,[578,1049,1514,2006,2363],{"id":579,"title":580,"alternatives":581,"api_compatible":25,"body":587,"category":504,"chinese_friendly":549,"cover":994,"description":995,"domestic":507,"extension":508,"faq":996,"free":507,"github":25,"languages":1009,"meta":1011,"models":25,"navigation":525,"notSuitable":25,"opensource":525,"path":21,"pillar":527,"platforms":1012,"priceTable":1014,"pricing":1022,"published":540,"relatedPlaybooks":1023,"relatedReviews":25,"score":1028,"self_host":525,"seo":1029,"slug":1030,"sources":1031,"stem":1038,"suitable":25,"tagline":1039,"tags":1040,"updated":555,"verdict":1046,"website":1047,"__hash__":1048},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Fcherry-studio.md","Cherry Studio",[582,583,584,586],{"name":23,"url":24},{"name":14,"url":15},{"name":585,"url":526},"ollama",{"name":17,"url":18},{"type":27,"value":588,"toc":982},[589,591,594,597,599,643,645,658,664,668,672,689,693,710,712,736,738,876,878,910,912,935,937,957,959],[30,590,33],{"id":32},[35,592,593],{},"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。",[35,595,596],{},"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。",[30,598,48],{"id":48},[50,600,601,607,613,619,625,631,637],{},[53,602,603,606],{},[56,604,605],{},"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",[53,608,609,612],{},[56,610,611],{},"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",[53,614,615,618],{},[56,616,617],{},"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",[53,620,621,624],{},[56,622,623],{},"MCP 协议","：扩展工具调用 \u002F 联网搜索 \u002F 文件操作",[53,626,627,630],{},[56,628,629],{},"数据本地优先","：对话历史本地存储，WebDAV 同步，不上传第三方",[53,632,633,636],{},[56,634,635],{},"多模态","：图片识别 \u002F PDF 阅读 \u002F Markdown + Mermaid + 代码高亮",[53,638,639,642],{},[56,640,641],{},"AI 绘画 + 翻译","：内置主流 SD \u002F DALL·E \u002F 翻译 API 集成",[30,644,124],{"id":124},[50,646,647,652],{},[53,648,649,651],{},[56,650,535],{},"：完全免费，AGPL-3.0",[53,653,654,657],{},[56,655,656],{},"Enterprise","：私有化部署 + 团队协作 + 资源管控，联系销售",[659,660,661],"blockquote",{},[35,662,663],{},"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。",[30,665,667],{"id":666},"实测mac-m2-中型知识库","实测（Mac M2 + 中型知识库）",[35,669,670],{},[56,671,136],{},[50,673,674,677,680,683,686],{},[53,675,676],{},"中文 UI \u002F 文档 \u002F 社区都顶级，零门槛上手",[53,678,679],{},"本地 RAG 拖入 30+ PDF 后向量化 \u003C 2 分钟（用 bge-m3）",[53,681,682],{},"多模型并排回答：让 Claude \u002F GPT \u002F DeepSeek 同回一个问题做比较",[53,684,685],{},"MCP 接 Brave Search + 自定义工具流畅",[53,687,688],{},"WebDAV 同步坚果云 \u002F 阿里云盘，桌面 + 移动设备数据互通",[35,690,691],{},[56,692,161],{},[50,694,695,698,701,704,707],{},[53,696,697],{},"没有 Web 端 \u002F Docker 自托管（要这个用 LobeChat）",[53,699,700],{},"iOS 版尚未发布（roadmap 中）",[53,702,703],{},"大型 PDF（>100 MB）向量化偶有失败，要切小",[53,705,706],{},"助手市场质量参差，要自筛",[53,708,709],{},"模型 API 调用全靠你自己付费，新手要先理解 API Key 概念",[30,711,195],{"id":195},[197,713,714,717,720,727,730,733],{},[53,715,716],{},"cherry-ai.com 下载客户端（或 GitHub releases）",[53,718,719],{},"设置 → 模型服务 → 填 OpenAI \u002F Claude \u002F DeepSeek API Key",[53,721,722,723],{},"（可选）本地：装 Ollama → Cherry Studio 自动识别 endpoint ",[443,724,725],{"href":725,"rel":726},"http:\u002F\u002Flocalhost:11434",[474],[53,728,729],{},"新建知识库 → 拖文件 \u002F 加网址 → 等向量化",[53,731,732],{},"新对话 → 选模型 → 勾知识库 → 提问",[53,734,735],{},"进阶：自定义助手（System Prompt）+ MCP 扩展工具",[30,737,233],{"id":233},[235,739,740,756],{},[238,741,742],{},[241,743,744,746,748,751,753],{},[244,745,246],{},[244,747,580],{},[244,749,750],{},"LobeChat",[244,752,251],{},[244,754,755],{},"Open WebUI",[259,757,758,773,787,802,815,830,845,860],{},[241,759,760,762,765,768,770],{},[264,761,266],{},[264,763,764],{},"桌面",[264,766,767],{},"Web + 桌面",[264,769,764],{},[264,771,772],{},"Docker \u002F 桌面",[241,774,775,777,780,782,785],{},[264,776,605],{},[264,778,779],{},"✅ 云 + 本地",[264,781,779],{},[264,783,784],{},"本地为主",[264,786,779],{},[241,788,789,792,795,797,800],{},[264,790,791],{},"知识库 RAG",[264,793,794],{},"✅ 强",[264,796,794],{},[264,798,799],{},"弱",[264,801,322],{},[241,803,804,807,809,811,813],{},[264,805,806],{},"MCP",[264,808,322],{},[264,810,322],{},[264,812,799],{},[264,814,322],{},[241,816,817,820,823,826,828],{},[264,818,819],{},"自托管 \u002F Web",[264,821,822],{},"无 Web",[264,824,825],{},"✅ Docker",[264,827,306],{},[264,829,825],{},[241,831,832,835,838,840,843],{},[264,833,834],{},"中文",[264,836,837],{},"5\u002F5",[264,839,837],{},[264,841,842],{},"4\u002F5",[264,844,842],{},[241,846,847,850,853,855,858],{},[264,848,849],{},"开源协议",[264,851,852],{},"AGPL-3.0",[264,854,364],{},[264,856,857],{},"闭源（免费）",[264,859,364],{},[241,861,862,865,868,871,873],{},[264,863,864],{},"GitHub Stars",[264,866,867],{},"60k+",[264,869,870],{},"72k+",[264,872,356],{},[264,874,875],{},"126k+",[30,877,375],{"id":375},[50,879,880,886,892,898,904],{},[53,881,882,885],{},[56,883,884],{},"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",[53,887,888,891],{},[56,889,890],{},"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",[53,893,894,897],{},[56,895,896],{},"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",[53,899,900,903],{},[56,901,902],{},"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",[53,905,906,909],{},[56,907,908],{},"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",[30,911,411],{"id":410},[50,913,914,917,920,923,926,929,932],{},[53,915,916],{},"✅ 中文用户、AI 重度使用 \u002F 多模型管理",[53,918,919],{},"✅ 需要本地 RAG 知识库",[53,921,922],{},"✅ 关注数据隐私 \u002F 本地存储",[53,924,925],{},"✅ 想用 Ollama \u002F LM Studio 本地模型",[53,927,928],{},"❌ 需要 Web 端 \u002F Docker 自托管",[53,930,931],{},"❌ 团队多人共享 \u002F SSO",[53,933,934],{},"❌ iOS 主力用户",[30,936,437],{"id":437},[50,938,939,944,948,953],{},[53,940,941],{},[443,942,943],{"href":24},"LobeChat 评测",[53,945,946],{},[443,947,445],{"href":15},[53,949,950],{},[443,951,952],{"href":526},"Ollama 评测",[53,954,955],{},[443,956,461],{"href":460},[30,958,464],{"id":464},[197,960,961,968,975],{},[53,962,963,964],{},"Cherry Studio 官网（功能 + 下载）",[443,965,966],{"href":966,"rel":967},"https:\u002F\u002Fwww.cherry-ai.com\u002F",[474],[53,969,970,971],{},"MBLUO Studio — Cherry Studio 评测 2026 ",[443,972,973],{"href":973,"rel":974},"https:\u002F\u002Fmbluostudio.com\u002Ftools\u002Fcherry-studio",[474],[53,976,977,978],{},"Cursor IDE 博客 — Cherry Studio 完全指南（2025-03）",[443,979,980],{"href":980,"rel":981},"https:\u002F\u002Fwww.cursor-ide.com\u002Fblog\u002Fcherry-studio-guide",[474],{"title":490,"searchDepth":491,"depth":491,"links":983},[984,985,986,987,988,989,990,991,992,993],{"id":32,"depth":494,"text":33},{"id":48,"depth":494,"text":48},{"id":124,"depth":494,"text":124},{"id":666,"depth":494,"text":667},{"id":195,"depth":494,"text":195},{"id":233,"depth":494,"text":233},{"id":375,"depth":494,"text":375},{"id":410,"depth":494,"text":411},{"id":437,"depth":494,"text":437},{"id":464,"depth":494,"text":464},"\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。",[997,1000,1003,1006],{"q":998,"a":999},"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":1001,"a":1002},"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":1004,"a":1005},"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":1007,"a":1008},"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。",[1010,523],"zh",{},[529,530,531,1013],"android",[1015,1018],{"plan":535,"price":536,"features":1016,"notes":1017},"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":656,"price":1019,"features":1020,"notes":1021},"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",[1024,1025],{"name":543,"url":460},{"name":1026,"url":1027},"Cursor MCP 深度集成","\u002Fplaybook\u002Fonboarding\u002Fcursor-mcp-deep-integration",{"power":548,"ux":549,"price":549,"cn_support":549,"stability":548},{"title":580,"description":995},"coding\u002Flocal\u002Fcherry-studio",[1032,1034,1036],{"name":1033,"url":966,"accessed":555},"Cherry Studio 官网",{"name":1035,"url":973,"accessed":555},"MBLUO Studio — Cherry Studio 评测",{"name":1037,"url":980,"accessed":555},"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",[504,1041,1042,1043,1044,570,1045],"desktop","multi-model","knowledge-base","rag","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":1050,"title":251,"alternatives":1051,"api_compatible":25,"body":1056,"category":504,"chinese_friendly":491,"cover":1466,"description":1467,"domestic":507,"extension":508,"faq":1468,"free":507,"github":25,"languages":1481,"meta":1482,"models":25,"navigation":525,"notSuitable":25,"opensource":507,"path":15,"pillar":527,"platforms":1483,"priceTable":1484,"pricing":1492,"published":540,"relatedPlaybooks":1493,"relatedReviews":25,"score":1496,"self_host":525,"seo":1497,"slug":1498,"sources":1499,"stem":1505,"suitable":25,"tagline":1506,"tags":1507,"updated":555,"verdict":1511,"website":1512,"__hash__":1513},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Flm-studio.md",[1052,1053,1054,1055],{"name":585,"url":526},{"name":17,"url":18},{"name":20,"url":21},{"name":23,"url":24},{"type":27,"value":1057,"toc":1454},[1058,1060,1067,1070,1072,1128,1130,1144,1149,1153,1157,1174,1178,1195,1197,1223,1225,1351,1353,1385,1387,1410,1412,1431,1433],[30,1059,33],{"id":32},[35,1061,1062,1063,1066],{},"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 + ",[39,1064,1065],{},"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。",[35,1068,1069],{},"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）。",[30,1071,48],{"id":48},[50,1073,1074,1080,1086,1092,1101,1110,1116,1122],{},[53,1075,1076,1079],{},[56,1077,1078],{},"模型浏览器","：内置 Hugging Face 检索，按 GGUF \u002F MLX \u002F 大小筛选、一键下载",[53,1081,1082,1085],{},[56,1083,1084],{},"聊天界面","：System Prompt \u002F temperature \u002F top-p \u002F context size 可视化调参",[53,1087,1088,1091],{},[56,1089,1090],{},"多模型并存 \u002F 切换","：同时加载多模型在不同会话中比较",[53,1093,1094,65,1097,1100],{},[56,1095,1096],{},"OpenAI 兼容 Local Server",[39,1098,1099],{},"http:\u002F\u002Flocalhost:1234\u002Fv1","，任何 SDK 即接即用",[53,1102,1103,65,1106,1109],{},[56,1104,1105],{},"Headless \u002F CLI",[39,1107,1108],{},"lms server start --port 1234","，无 GUI 可跑",[53,1111,1112,1115],{},[56,1113,1114],{},"PDF \u002F 文档对话","：内置基础 RAG，丢文件就能聊",[53,1117,1118,1121],{},[56,1119,1120],{},"MLX 原生支持（Mac）","：M1+ 上比 GGUF + Metal 快 30–50%",[53,1123,1124,1127],{},[56,1125,1126],{},"持续批处理","：Codersera 2026 测得 50–90 tok\u002Fs（消费级 GPU + 中等模型）",[30,1129,124],{"id":124},[50,1131,1132,1138],{},[53,1133,1134,1137],{},[56,1135,1136],{},"个人 \u002F 评估","：免费，全功能可用",[53,1139,1140,1143],{},[56,1141,1142],{},"商用","：邮件 \u002F 官网联系 LM Studio 团队",[659,1145,1146],{},[35,1147,1148],{},"模型本身免费（开源权重），LM Studio 不抽水任何 token 费用。",[30,1150,1152],{"id":1151},"实测mac-m2-pro-qwen3-coder-7b-gguf-q4_k_m","实测（Mac M2 Pro + Qwen3-Coder-7B GGUF Q4_K_M）",[35,1154,1155],{},[56,1156,136],{},[50,1158,1159,1162,1165,1168,1171],{},[53,1160,1161],{},"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",[53,1163,1164],{},"加载 7B Q4 模型 \u003C 3 秒，生成 ~75 tok\u002Fs",[53,1166,1167],{},"Local Server 开了 Cursor 直接接 baseURL → 本地代码补全零成本",[53,1169,1170],{},"MLX 版同模型 ~110 tok\u002Fs，差距显著",[53,1172,1173],{},"多窗口加载 2 个模型并排测，调 prompt 直观",[35,1175,1176],{},[56,1177,161],{},[50,1179,1180,1183,1186,1189,1192],{},[53,1181,1182],{},"模型库依赖 Hugging Face，国内访问要镜像 \u002F 代理",[53,1184,1185],{},"GPU 显存吃满后会自动 offload 到 CPU，无提示就慢下来",[53,1187,1188],{},"Headless 模式相对 Ollama 偏新，文档稍少",[53,1190,1191],{},"闭源应用（虽免费），不适合企业合规挂钩",[53,1193,1194],{},"中文 UI 可用但部分菜单仍英文",[30,1196,195],{"id":195},[197,1198,1199,1202,1205,1208,1211,1218],{},[53,1200,1201],{},"lmstudio.ai 下载（Mac \u002F Windows \u002F Linux）",[53,1203,1204],{},"打开 → Discover 标签 → 搜模型（如 qwen3-coder、deepseek-v3 GGUF\u002FMLX）→ Download",[53,1206,1207],{},"Chat 标签 → 选模型 → 调参聊天",[53,1209,1210],{},"Local Server 标签 → Start Server → 默认端口 1234",[53,1212,1213,1214,1217],{},"在你的应用里：",[39,1215,1216],{},"baseURL = \"http:\u002F\u002Flocalhost:1234\u002Fv1\"","，API Key 任意",[53,1219,1220,1221],{},"Headless：",[39,1222,1108],{},[30,1224,233],{"id":233},[235,1226,1227,1241],{},[238,1228,1229],{},[241,1230,1231,1233,1235,1237,1239],{},[244,1232,246],{},[244,1234,251],{},[244,1236,11],{},[244,1238,755],{},[244,1240,257],{},[259,1242,1243,1259,1274,1288,1300,1312,1325,1337],{},[241,1244,1245,1247,1250,1253,1256],{},[264,1246,266],{},[264,1248,1249],{},"GUI + CLI",[264,1251,1252],{},"CLI Daemon",[264,1254,1255],{},"Docker UI",[264,1257,1258],{},"二进制",[241,1260,1261,1263,1266,1269,1271],{},[264,1262,298],{},[264,1264,1265],{},"✅ 内置",[264,1267,1268],{},"CLI pull",[264,1270,306],{},[264,1272,1273],{},"手动",[241,1275,1276,1279,1281,1284,1286],{},[264,1277,1278],{},"参数调优 GUI",[264,1280,322],{},[264,1282,1283],{},"❌",[264,1285,353],{},[264,1287,1283],{},[241,1289,1290,1292,1294,1296,1298],{},[264,1291,79],{},[264,1293,319],{},[264,1295,316],{},[264,1297,322],{},[264,1299,322],{},[241,1301,1302,1304,1306,1308,1310],{},[264,1303,345],{},[264,1305,322],{},[264,1307,348],{},[264,1309,356],{},[264,1311,356],{},[241,1313,1314,1317,1319,1321,1323],{},[264,1315,1316],{},"多用户并发",[264,1318,799],{},[264,1320,799],{},[264,1322,322],{},[264,1324,290],{},[241,1326,1327,1329,1331,1333,1335],{},[264,1328,361],{},[264,1330,857],{},[264,1332,364],{},[264,1334,364],{},[264,1336,364],{},[241,1338,1339,1342,1344,1347,1349],{},[264,1340,1341],{},"上手难度",[264,1343,285],{},[264,1345,1346],{},"低",[264,1348,290],{},[264,1350,293],{},[30,1352,375],{"id":375},[50,1354,1355,1361,1367,1373,1379],{},[53,1356,1357,1360],{},[56,1358,1359],{},"国内下模型走镜像","：HF 直连慢 \u002F 卡，配 HF_ENDPOINT=hf-mirror.com",[53,1362,1363,1366],{},[56,1364,1365],{},"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",[53,1368,1369,1372],{},[56,1370,1371],{},"MLX 优先（Mac M 系列）","：能下 MLX 版就别下 GGUF，速度差距明显",[53,1374,1375,1378],{},[56,1376,1377],{},"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",[53,1380,1381,1384],{},[56,1382,1383],{},"闭源合规要核","：企业内部使用前查 license；商用必须联系官方",[30,1386,411],{"id":410},[50,1388,1389,1392,1395,1398,1401,1404,1407],{},[53,1390,1391],{},"✅ 本地 LLM 入门 \u002F 评估",[53,1393,1394],{},"✅ Mac M 系列用户",[53,1396,1397],{},"✅ 想给 Cursor \u002F Cline 接本地 OpenAI 兼容 endpoint",[53,1399,1400],{},"✅ 需要 GUI 调参 \u002F 模型比较",[53,1402,1403],{},"❌ 多用户并发生产服务",[53,1405,1406],{},"❌ 嵌入式 \u002F 边缘设备",[53,1408,1409],{},"❌ 强合规 \u002F 必须开源审计",[30,1411,437],{"id":437},[50,1413,1414,1418,1422,1426],{},[53,1415,1416],{},[443,1417,952],{"href":526},[53,1419,1420],{},[443,1421,450],{"href":18},[53,1423,1424],{},[443,1425,455],{"href":21},[53,1427,1428],{},[443,1429,1430],{"href":546},"Claude Code 上手 Playbook",[30,1432,464],{"id":464},[197,1434,1435,1442,1449],{},[53,1436,1437,1438],{},"LM Studio 官网 ",[443,1439,1440],{"href":1440,"rel":1441},"https:\u002F\u002Flmstudio.ai\u002F",[474],[53,1443,1444,1445],{},"Codersera — LM Studio Complete Guide 2026 ",[443,1446,1447],{"href":1447,"rel":1448},"https:\u002F\u002Fcodersera.com\u002Fblog\u002Flm-studio-complete-guide-2026\u002F",[474],[53,1450,484,1451],{},[443,1452,487],{"href":487,"rel":1453},[474],{"title":490,"searchDepth":491,"depth":491,"links":1455},[1456,1457,1458,1459,1460,1461,1462,1463,1464,1465],{"id":32,"depth":494,"text":33},{"id":48,"depth":494,"text":48},{"id":124,"depth":494,"text":124},{"id":1151,"depth":494,"text":1152},{"id":195,"depth":494,"text":195},{"id":233,"depth":494,"text":233},{"id":375,"depth":494,"text":375},{"id":410,"depth":494,"text":411},{"id":437,"depth":494,"text":437},{"id":464,"depth":494,"text":464},"\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。",[1469,1472,1475,1478],{"q":1470,"a":1471},"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":1473,"a":1474},"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":1476,"a":1477},"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":1479,"a":1480},"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。",[523,1010],{},[529,530,531],[1485,1488],{"plan":1136,"price":536,"features":1486,"notes":1487},"全功能 GUI + Headless API + GGUF\u002FMLX","供个人 \u002F 评估使用",{"plan":1142,"price":1489,"features":1490,"notes":1491},"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",[1494,1495],{"name":543,"url":460},{"name":545,"url":546},{"power":548,"ux":549,"price":549,"cn_support":491,"stability":548},{"title":251,"description":1467},"coding\u002Flocal\u002Flm-studio",[1500,1502,1504],{"name":1501,"url":1440,"accessed":555},"LM Studio 官网",{"name":1503,"url":1447,"accessed":555},"Codersera — LM Studio Complete Guide 2026",{"name":559,"url":487,"accessed":555},"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",[504,1508,567,568,1509,1510,569],"gui","llama-cpp","mac","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":1515,"title":750,"alternatives":1516,"api_compatible":25,"body":1521,"category":504,"chinese_friendly":549,"cover":1957,"description":1958,"domestic":507,"extension":508,"faq":1959,"free":507,"github":25,"languages":1972,"meta":1973,"models":25,"navigation":525,"notSuitable":25,"opensource":525,"path":24,"pillar":527,"platforms":1974,"priceTable":1976,"pricing":1984,"published":540,"relatedPlaybooks":1985,"relatedReviews":25,"score":1988,"self_host":525,"seo":1989,"slug":1990,"sources":1991,"stem":1998,"suitable":25,"tagline":1999,"tags":2000,"updated":555,"verdict":2003,"website":2004,"__hash__":2005},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Flobe-chat.md",[1517,1518,1519,1520],{"name":20,"url":21},{"name":17,"url":18},{"name":585,"url":526},{"name":14,"url":15},{"type":27,"value":1522,"toc":1945},[1523,1525,1532,1535,1537,1599,1601,1614,1617,1621,1625,1645,1649,1666,1668,1694,1696,1832,1834,1872,1874,1900,1902,1920,1922],[30,1524,33],{"id":32},[35,1526,1527,1528,1531],{},"LobeChat 是 LobeHub 团队的开源 AI 聊天框架，2023 年发布、GitHub 72k+ stars、MIT 协议。",[56,1529,1530],{},"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。",[35,1533,1534],{},"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）。",[30,1536,48],{"id":48},[50,1538,1539,1544,1550,1555,1561,1567,1572,1577,1583,1589],{},[53,1540,1541,1543],{},[56,1542,605],{},"：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",[53,1545,1546,1549],{},[56,1547,1548],{},"多模型对比","：同 prompt 给多模型并排回答",[53,1551,1552,1554],{},[56,1553,611],{},"：上传 PDF \u002F Word \u002F 网页 → 向量化 → 检索引用",[53,1556,1557,1560],{},[56,1558,1559],{},"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",[53,1562,1563,1566],{},[56,1564,1565],{},"助手市场","：几百个预设 AI 角色，一键导入",[53,1568,1569,1571],{},[56,1570,623],{},"：扩展任意工具能力",[53,1573,1574],{},[56,1575,1576],{},"代码解释器 \u002F 文件上传 \u002F TTS \u002F 多模态",[53,1578,1579,1582],{},[56,1580,1581],{},"Web + 桌面 + Docker","：三形态，数据可完全本地",[53,1584,1585,1588],{},[56,1586,1587],{},"LobeHub Cloud","：官方云托管，免部署",[53,1590,1591,1594,1595,1598],{},[56,1592,1593],{},"快捷指令 \u002F 工作流","：自定义 prompt 模板，",[39,1596,1597],{},"\u002Fpodcast-summary"," 类用法",[30,1600,124],{"id":124},[50,1602,1603,1609],{},[53,1604,1605,1608],{},[56,1606,1607],{},"自托管 \u002F 桌面","：完全免费、MIT 开源",[53,1610,1611,1613],{},[56,1612,1587],{},"：订阅制，云端托管 + 团队协作 + 同步",[35,1615,1616],{},"模型 API 费用按你自己的供应商付费；本地 Ollama \u002F LM Studio 零成本。",[30,1618,1620],{"id":1619},"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）",[35,1622,1623],{},[56,1624,136],{},[50,1626,1627,1630,1633,1636,1639,1642],{},[53,1628,1629],{},"界面颜值是这一类工具里第一档（深色 \u002F 透明 \u002F 现代感）",[53,1631,1632],{},"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",[53,1634,1635],{},"知识库 RAG 上传 50+ PDF 后检索准确，引用片段可视化",[53,1637,1638],{},"助手市场拿来即用——「Code Reviewer」「Translation Polish」节省 prompt 编写",[53,1640,1641],{},"Docker 一键部署，团队 5 人共享流畅",[53,1643,1644],{},"多平台数据同步（Cloud \u002F WebDAV）",[35,1646,1647],{},[56,1648,161],{},[50,1650,1651,1654,1657,1660,1663],{},[53,1652,1653],{},"自托管要熟悉 Docker + 反代 + HTTPS",[53,1655,1656],{},"国内连 OpenAI \u002F Claude 需自带网络方案",[53,1658,1659],{},"Web 版数据存 LobeHub，隐私敏感场景走桌面 \u002F Docker",[53,1661,1662],{},"插件市场质量参差，要自筛",[53,1664,1665],{},"团队多人共享需配 LobeHub Cloud 或自建数据库（Postgres + S3）",[30,1667,195],{"id":195},[197,1669,1670,1673,1679,1682,1685,1688,1691],{},[53,1671,1672],{},"选形态：Web（chat.lobehub.com 注册即用） \u002F 桌面（GitHub Releases 下载） \u002F Docker",[53,1674,1675,1676],{},"Docker：",[39,1677,1678],{},"docker run -d -p 3210:3210 -e OPENAI_API_KEY=sk-xxx --name lobe-chat lobehub\u002Flobe-chat",[53,1680,1681],{},"设置 → AI 服务商 → 添加 OpenAI \u002F Claude \u002F DeepSeek \u002F Ollama",[53,1683,1684],{},"模型选择器测试对话",[53,1686,1687],{},"知识库：拖文件 → 等向量化 → 对话引用",[53,1689,1690],{},"助手市场拉「Code Reviewer」「论文翻译润色」试用",[53,1692,1693],{},"进阶：插件市场启用联网搜索 \u002F 代码执行；MCP 自定义工具",[30,1695,233],{"id":233},[235,1697,1698,1712],{},[238,1699,1700],{},[241,1701,1702,1704,1706,1708,1710],{},[244,1703,246],{},[244,1705,750],{},[244,1707,580],{},[244,1709,755],{},[244,1711,251],{},[259,1713,1714,1726,1739,1752,1765,1781,1794,1808,1820],{},[241,1715,1716,1718,1720,1722,1724],{},[264,1717,266],{},[264,1719,1581],{},[264,1721,764],{},[264,1723,772],{},[264,1725,764],{},[241,1727,1728,1730,1733,1735,1737],{},[264,1729,605],{},[264,1731,1732],{},"✅ 80+",[264,1734,322],{},[264,1736,322],{},[264,1738,784],{},[241,1740,1741,1743,1746,1748,1750],{},[264,1742,1548],{},[264,1744,1745],{},"✅ 一等",[264,1747,322],{},[264,1749,799],{},[264,1751,799],{},[241,1753,1754,1756,1758,1760,1763],{},[264,1755,791],{},[264,1757,322],{},[264,1759,322],{},[264,1761,1762],{},"✅ + oikb",[264,1764,799],{},[241,1766,1767,1770,1773,1776,1779],{},[264,1768,1769],{},"插件 \u002F 助手市场",[264,1771,1772],{},"✅ 丰富",[264,1774,1775],{},"300+ 助手",[264,1777,1778],{},"Tools",[264,1780,799],{},[241,1782,1783,1785,1787,1789,1792],{},[264,1784,806],{},[264,1786,322],{},[264,1788,322],{},[264,1790,1791],{},"✅ mcpo",[264,1793,799],{},[241,1795,1796,1799,1802,1804,1806],{},[264,1797,1798],{},"多用户",[264,1800,1801],{},"配 Cloud \u002F 自建",[264,1803,306],{},[264,1805,1745],{},[264,1807,306],{},[241,1809,1810,1812,1814,1816,1818],{},[264,1811,864],{},[264,1813,870],{},[264,1815,867],{},[264,1817,875],{},[264,1819,356],{},[241,1821,1822,1824,1826,1828,1830],{},[264,1823,849],{},[264,1825,364],{},[264,1827,852],{},[264,1829,364],{},[264,1831,857],{},[30,1833,375],{"id":375},[50,1835,1836,1842,1848,1854,1860,1866],{},[53,1837,1838,1841],{},[56,1839,1840],{},"Web 版数据不本地","：隐私敏感选桌面或 Docker 自托管",[53,1843,1844,1847],{},[56,1845,1846],{},"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",[53,1849,1850,1853],{},[56,1851,1852],{},"Docker 自托管暴露公网","：上反代 + HTTPS + Auth + 备份数据库",[53,1855,1856,1859],{},[56,1857,1858],{},"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",[53,1861,1862,1865],{},[56,1863,1864],{},"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",[53,1867,1868,1871],{},[56,1869,1870],{},"同步选 Cloud vs WebDAV","：团队多端走 LobeHub Cloud；个人多设备 WebDAV 即可",[30,1873,411],{"id":410},[50,1875,1876,1879,1882,1885,1888,1891,1894,1897],{},[53,1877,1878],{},"✅ Web + 桌面双形态需求",[53,1880,1881],{},"✅ Docker 自托管 \u002F 团队共享",[53,1883,1884],{},"✅ 多模型对比 \u002F 选型",[53,1886,1887],{},"✅ 中文重度用户",[53,1889,1890],{},"✅ 助手市场 \u002F 插件生态用户",[53,1892,1893],{},"❌ 强企业 RBAC + 多租户（Open WebUI 更完善）",[53,1895,1896],{},"❌ 只要桌面 + 数据完全本地（Cherry Studio 同样优秀）",[53,1898,1899],{},"❌ 完全不会碰 Docker",[30,1901,437],{"id":437},[50,1903,1904,1908,1912,1916],{},[53,1905,1906],{},[443,1907,455],{"href":21},[53,1909,1910],{},[443,1911,450],{"href":18},[53,1913,1914],{},[443,1915,952],{"href":526},[53,1917,1918],{},[443,1919,461],{"href":460},[30,1921,464],{"id":464},[197,1923,1924,1931,1938],{},[53,1925,1926,1927],{},"LobeChat GitHub 仓库（72k+ stars，MIT）",[443,1928,1929],{"href":1929,"rel":1930},"https:\u002F\u002Fgithub.com\u002Flobehub\u002Flobe-chat",[474],[53,1932,1933,1934],{},"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）",[443,1935,1936],{"href":1936,"rel":1937},"https:\u002F\u002Fcloud.tencent.com\u002Fdeveloper\u002Farticle\u002F2622150",[474],[53,1939,1940,1941],{},"Ofox.ai — LobeChat 完全配置指南 2026（2026-04-17）",[443,1942,1943],{"href":1943,"rel":1944},"https:\u002F\u002Fofox.ai\u002Fzh\u002Fblog\u002Flobechat-api-configuration-guide-2026",[474],{"title":490,"searchDepth":491,"depth":491,"links":1946},[1947,1948,1949,1950,1951,1952,1953,1954,1955,1956],{"id":32,"depth":494,"text":33},{"id":48,"depth":494,"text":48},{"id":124,"depth":494,"text":124},{"id":1619,"depth":494,"text":1620},{"id":195,"depth":494,"text":195},{"id":233,"depth":494,"text":233},{"id":375,"depth":494,"text":375},{"id":410,"depth":494,"text":411},{"id":437,"depth":494,"text":437},{"id":464,"depth":494,"text":464},"\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。",[1960,1963,1966,1969],{"q":1961,"a":1962},"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":1964,"a":1965},"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":1967,"a":1968},"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":1970,"a":1971},"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。",[1010,523],{},[1975,529,530,531,532],"web",[1977,1980],{"plan":1607,"price":536,"features":1978,"notes":1979},"全功能 \u002F 80+ 模型 \u002F 知识库 \u002F 插件 \u002F 助手市场","MIT 协议",{"plan":1587,"price":1981,"features":1982,"notes":1983},"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",[1986,1987],{"name":543,"url":460},{"name":545,"url":546},{"power":549,"ux":549,"price":549,"cn_support":549,"stability":548},{"title":750,"description":1958},"coding\u002Flocal\u002Flobe-chat",[1992,1994,1996],{"name":1993,"url":1929,"accessed":555},"LobeChat GitHub",{"name":1995,"url":1936,"accessed":555},"腾讯云开发者社区 — Lobe Chat 终极桌面客户端",{"name":1997,"url":1943,"accessed":555},"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",[504,1975,1041,1042,1044,2001,2002,570],"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":10,"title":11,"alternatives":2007,"api_compatible":25,"body":2012,"category":504,"chinese_friendly":491,"cover":505,"description":506,"domestic":507,"extension":508,"faq":2343,"free":507,"github":25,"languages":2348,"meta":2349,"models":25,"navigation":525,"notSuitable":25,"opensource":525,"path":526,"pillar":527,"platforms":2350,"priceTable":2351,"pricing":539,"published":540,"relatedPlaybooks":2353,"relatedReviews":25,"score":2356,"self_host":525,"seo":2357,"slug":551,"sources":2358,"stem":560,"suitable":25,"tagline":561,"tags":2362,"updated":555,"verdict":571,"website":572,"__hash__":573},[2008,2009,2010,2011],{"name":14,"url":15},{"name":17,"url":18},{"name":20,"url":21},{"name":23,"url":24},{"type":27,"value":2013,"toc":2331},[2014,2016,2020,2022,2024,2072,2074,2076,2078,2082,2096,2100,2120,2122,2146,2148,2250,2252,2274,2276,2292,2294,2312,2314],[30,2015,33],{"id":32},[35,2017,37,2018,42],{},[39,2019,41],{},[35,2021,45],{},[30,2023,48],{"id":48},[50,2025,2026,2030,2036,2040,2046,2056,2060,2064,2068],{},[53,2027,2028,59],{},[56,2029,58],{},[53,2031,2032,65,2034],{},[56,2033,64],{},[39,2035,68],{},[53,2037,2038,74],{},[56,2039,73],{},[53,2041,2042,65,2044],{},[56,2043,79],{},[39,2045,82],{},[53,2047,2048,65,2050,91,2052,91,2054],{},[56,2049,87],{},[39,2051,90],{},[39,2053,94],{},[39,2055,97],{},[53,2057,2058,103],{},[56,2059,102],{},[53,2061,2062,109],{},[56,2063,108],{},[53,2065,2066,115],{},[56,2067,114],{},[53,2069,2070,121],{},[56,2071,120],{},[30,2073,124],{"id":124},[35,2075,127],{},[30,2077,131],{"id":130},[35,2079,2080],{},[56,2081,136],{},[50,2083,2084,2088,2090,2092,2094],{},[53,2085,2086,144],{},[39,2087,143],{},[53,2089,147],{},[53,2091,150],{},[53,2093,153],{},[53,2095,156],{},[35,2097,2098],{},[56,2099,161],{},[50,2101,2102,2108,2112,2116,2118],{},[53,2103,166,2104,170,2106],{},[39,2105,169],{},[39,2107,173],{},[53,2109,176,2110],{},[39,2111,179],{},[53,2113,182,2114,186],{},[39,2115,185],{},[53,2117,189],{},[53,2119,192],{},[30,2121,195],{"id":195},[197,2123,2124,2128,2132,2136,2140,2144],{},[53,2125,2126,204],{},[39,2127,203],{},[53,2129,2130,210],{},[39,2131,209],{},[53,2133,2134,215],{},[39,2135,143],{},[53,2137,218,2138],{},[39,2139,221],{},[53,2141,224,2142],{},[39,2143,227],{},[53,2145,230],{},[30,2147,233],{"id":233},[235,2149,2150,2164],{},[238,2151,2152],{},[241,2153,2154,2156,2158,2160,2162],{},[244,2155,246],{},[244,2157,11],{},[244,2159,251],{},[244,2161,254],{},[244,2163,257],{},[259,2165,2166,2178,2190,2202,2214,2226,2238],{},[241,2167,2168,2170,2172,2174,2176],{},[264,2169,266],{},[264,2171,269],{},[264,2173,272],{},[264,2175,275],{},[264,2177,278],{},[241,2179,2180,2182,2184,2186,2188],{},[264,2181,195],{},[264,2183,285],{},[264,2185,285],{},[264,2187,290],{},[264,2189,293],{},[241,2191,2192,2194,2196,2198,2200],{},[264,2193,298],{},[264,2195,64],{},[264,2197,303],{},[264,2199,306],{},[264,2201,306],{},[241,2203,2204,2206,2208,2210,2212],{},[264,2205,313],{},[264,2207,316],{},[264,2209,319],{},[264,2211,322],{},[264,2213,322],{},[241,2215,2216,2218,2220,2222,2224],{},[264,2217,329],{},[264,2219,332],{},[264,2221,335],{},[264,2223,338],{},[264,2225,290],{},[241,2227,2228,2230,2232,2234,2236],{},[264,2229,345],{},[264,2231,348],{},[264,2233,322],{},[264,2235,353],{},[264,2237,356],{},[241,2239,2240,2242,2244,2246,2248],{},[264,2241,361],{},[264,2243,364],{},[264,2245,367],{},[264,2247,370],{},[264,2249,364],{},[30,2251,375],{"id":375},[50,2253,2254,2258,2262,2266,2270],{},[53,2255,2256,383],{},[56,2257,382],{},[53,2259,2260,389],{},[56,2261,388],{},[53,2263,2264,395],{},[56,2265,394],{},[53,2267,2268,401],{},[56,2269,400],{},[53,2271,2272,407],{},[56,2273,406],{},[30,2275,411],{"id":410},[50,2277,2278,2280,2282,2284,2286,2288,2290],{},[53,2279,416],{},[53,2281,419],{},[53,2283,422],{},[53,2285,425],{},[53,2287,428],{},[53,2289,431],{},[53,2291,434],{},[30,2293,437],{"id":437},[50,2295,2296,2300,2304,2308],{},[53,2297,2298],{},[443,2299,445],{"href":15},[53,2301,2302],{},[443,2303,450],{"href":18},[53,2305,2306],{},[443,2307,455],{"href":21},[53,2309,2310],{},[443,2311,461],{"href":460},[30,2313,464],{"id":464},[197,2315,2316,2321,2326],{},[53,2317,469,2318],{},[443,2319,472],{"href":472,"rel":2320},[474],[53,2322,477,2323],{},[443,2324,480],{"href":480,"rel":2325},[474],[53,2327,484,2328],{},[443,2329,487],{"href":487,"rel":2330},[474],{"title":490,"searchDepth":491,"depth":491,"links":2332},[2333,2334,2335,2336,2337,2338,2339,2340,2341,2342],{"id":32,"depth":494,"text":33},{"id":48,"depth":494,"text":48},{"id":124,"depth":494,"text":124},{"id":130,"depth":494,"text":131},{"id":195,"depth":494,"text":195},{"id":233,"depth":494,"text":233},{"id":375,"depth":494,"text":375},{"id":410,"depth":494,"text":411},{"id":437,"depth":494,"text":437},{"id":464,"depth":494,"text":464},[2344,2345,2346,2347],{"q":511,"a":512},{"q":514,"a":515},{"q":517,"a":518},{"q":520,"a":521},[523],{},[529,530,531,532],[2352],{"plan":535,"price":536,"features":537,"notes":538},[2354,2355],{"name":543,"url":460},{"name":545,"url":546},{"power":548,"ux":548,"price":549,"cn_support":491,"stability":549},{"title":11,"description":506},[2359,2360,2361],{"name":554,"url":472,"accessed":555},{"name":557,"url":480,"accessed":555},{"name":559,"url":487,"accessed":555},[504,563,564,565,566,567,568,569,570],{"id":2364,"title":755,"alternatives":2365,"api_compatible":25,"body":2370,"category":504,"chinese_friendly":548,"cover":2785,"description":2786,"domestic":507,"extension":508,"faq":2787,"free":507,"github":25,"languages":2800,"meta":2801,"models":25,"navigation":525,"notSuitable":25,"opensource":525,"path":18,"pillar":527,"platforms":2802,"priceTable":2804,"pricing":2811,"published":540,"relatedPlaybooks":2812,"relatedReviews":25,"score":2815,"self_host":525,"seo":2816,"slug":2817,"sources":2818,"stem":2825,"suitable":25,"tagline":2826,"tags":2827,"updated":555,"verdict":2830,"website":2831,"__hash__":2832},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Fopen-webui.md",[2366,2367,2368,2369],{"name":23,"url":24},{"name":20,"url":21},{"name":585,"url":526},{"name":14,"url":15},{"type":27,"value":2371,"toc":2773},[2372,2374,2377,2380,2382,2444,2446,2449,2453,2457,2485,2489,2513,2515,2549,2551,2658,2660,2703,2705,2728,2730,2748,2750],[30,2373,33],{"id":32},[35,2375,2376],{},"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。",[35,2378,2379],{},"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。",[30,2381,48],{"id":48},[50,2383,2384,2390,2396,2402,2408,2414,2420,2426,2432,2438],{},[53,2385,2386,2389],{},[56,2387,2388],{},"多模型后端","：Ollama \u002F OpenAI \u002F vLLM \u002F Anthropic \u002F Groq \u002F LocalAI \u002F 任意 OpenAI 兼容",[53,2391,2392,2395],{},[56,2393,2394],{},"多用户 + RBAC","：注册 \u002F 邀请 \u002F 角色权限 \u002F 工作区隔离",[53,2397,2398,2401],{},[56,2399,2400],{},"RAG 知识库","：上传文档 \u002F 网址 \u002F SearXNG 联网搜索 → 向量化 → 对话引用",[53,2403,2404,2407],{},[56,2405,2406],{},"Tools \u002F Functions","：Python 写函数即扩展（联网 \u002F 计算器 \u002F 自定义 API）",[53,2409,2410,2413],{},[56,2411,2412],{},"mcpo","：MCP-to-OpenAPI 代理，任意 MCP 服务器接进来",[53,2415,2416,2419],{},[56,2417,2418],{},"oikb","：知识库同步本地文件夹 \u002F GitHub \u002F S3 \u002F Confluence 等 40+ 源",[53,2421,2422,2425],{},[56,2423,2424],{},"open-terminal \u002F cptr","：给 AI 真实终端 + 文件 + 沙箱执行",[53,2427,2428,2431],{},[56,2429,2430],{},"图像生成","：Stable Diffusion \u002F DALL·E \u002F 自托管接入",[53,2433,2434,2437],{},[56,2435,2436],{},"语音输入 \u002F TTS","：内置",[53,2439,2440,2443],{},[56,2441,2442],{},"企业 LTS","：custom branding + SLA + 长期支持版本（联系销售）",[30,2445,124],{"id":124},[35,2447,2448],{},"完全免费、MIT 开源、商用免费。Enterprise 提供品牌定制 + SLA + LTS。",[30,2450,2452],{"id":2451},"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）",[35,2454,2455],{},[56,2456,136],{},[50,2458,2459,2466,2469,2476,2479,2482],{},[53,2460,2461,2462,2465],{},"单条 ",[39,2463,2464],{},"docker run"," 五分钟上线",[53,2467,2468],{},"自带的多用户 + 角色权限省去重新搭 Auth",[53,2470,2471,2472,2475],{},"RAG 直传 30 个 PDF 后向量化顺利，对话中 ",[39,2473,2474],{},"#知识库"," 引用准确",[53,2477,2478],{},"mcpo 把 GitHub MCP 服务器接进来，团队对话里直接 issue \u002F PR 操作",[53,2480,2481],{},"模型切换流畅，OpenAI + Ollama 并存",[53,2483,2484],{},"SearXNG 联网搜索给模型实时信息，过时知识截止问题缓解",[35,2486,2487],{},[56,2488,161],{},[50,2490,2491,2494,2500,2507,2510],{},[53,2492,2493],{},"Docker 镜像 ~1.5GB，首次拉取偏慢",[53,2495,166,2496,2499],{},[39,2497,2498],{},"0.0.0.0"," 公网暴露要加 HTTPS + 反代",[53,2501,2502,2503,2506],{},"嵌入模型 ",[39,2504,2505],{},"sentence-transformers"," 中文效果一般，建议换 bge-m3",[53,2508,2509],{},"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）",[53,2511,2512],{},"版本升级要看 changelog，部分 minor 含 breaking 改动",[30,2514,195],{"id":195},[197,2516,2517,2523,2530,2533,2536,2539,2542],{},[53,2518,2519,2520],{},"装 Docker → ",[39,2521,2522],{},"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",[53,2524,2525,2526,2529],{},"浏览器开 ",[39,2527,2528],{},"http:\u002F\u002Flocalhost:3000"," → 注册第一个账号（管理员）",[53,2531,2532],{},"设置 → Connections → 连接 Ollama \u002F 加 OpenAI Key",[53,2534,2535],{},"Models → Pull \u002F Discover 模型",[53,2537,2538],{},"Workspaces → 建知识库 → 上传文档",[53,2540,2541],{},"Tools → 启用 \u002F 写自定义函数",[53,2543,2544,2545,2548],{},"生产部署：Nginx 反代 + Let's Encrypt + 备份 ",[39,2546,2547],{},"\u002Fapp\u002Fbackend\u002Fdata"," volume",[30,2550,233],{"id":233},[235,2552,2553,2567],{},[238,2554,2555],{},[241,2556,2557,2559,2561,2563,2565],{},[244,2558,246],{},[244,2560,755],{},[244,2562,750],{},[244,2564,580],{},[244,2566,251],{},[259,2568,2569,2581,2593,2607,2620,2634,2646],{},[241,2570,2571,2573,2575,2577,2579],{},[264,2572,266],{},[264,2574,772],{},[264,2576,767],{},[264,2578,764],{},[264,2580,764],{},[241,2582,2583,2585,2587,2589,2591],{},[264,2584,1798],{},[264,2586,1745],{},[264,2588,322],{},[264,2590,306],{},[264,2592,306],{},[241,2594,2595,2598,2601,2603,2605],{},[264,2596,2597],{},"RAG",[264,2599,2600],{},"✅ 强 + oikb",[264,2602,322],{},[264,2604,322],{},[264,2606,799],{},[241,2608,2609,2612,2614,2616,2618],{},[264,2610,2611],{},"工具 \u002F MCP",[264,2613,1791],{},[264,2615,322],{},[264,2617,322],{},[264,2619,799],{},[241,2621,2622,2625,2628,2630,2632],{},[264,2623,2624],{},"自托管",[264,2626,2627],{},"✅ Docker \u002F K8s",[264,2629,825],{},[264,2631,306],{},[264,2633,306],{},[241,2635,2636,2638,2640,2642,2644],{},[264,2637,864],{},[264,2639,875],{},[264,2641,870],{},[264,2643,867],{},[264,2645,356],{},[241,2647,2648,2650,2652,2654,2656],{},[264,2649,849],{},[264,2651,364],{},[264,2653,364],{},[264,2655,852],{},[264,2657,857],{},[30,2659,375],{"id":375},[50,2661,2662,2668,2676,2685,2691,2697],{},[53,2663,2664,2667],{},[56,2665,2666],{},"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",[53,2669,2670,2675],{},[56,2671,2672,2673,2548],{},"备份 ",[39,2674,2547],{},"：知识库 \u002F 用户 \u002F 对话全在里面",[53,2677,2678,2681,2682,2684],{},[56,2679,2680],{},"中文 RAG 换嵌入模型","：默认 ",[39,2683,2505],{}," 中文一般，配 bge-m3 或硅基流动嵌入 API",[53,2686,2687,2690],{},[56,2688,2689],{},"mcpo 工具范围谨慎","：MCP 给 AI 真实能力，第三方服务器审一遍",[53,2692,2693,2696],{},[56,2694,2695],{},"后端吞吐看 Ollama","：5+ 并发上 vLLM 后端，Ollama 单 worker 会排队",[53,2698,2699,2702],{},[56,2700,2701],{},"升级前看 changelog","：weekly 更新，偶有 breaking",[30,2704,411],{"id":410},[50,2706,2707,2710,2713,2716,2719,2722,2725],{},[53,2708,2709],{},"✅ 团队 \u002F 家庭 \u002F 公司多人共享 AI 平台",[53,2711,2712],{},"✅ 要 Web 端访问 \u002F 移动端兼容",[53,2714,2715],{},"✅ 自托管 \u002F 完全控制数据",[53,2717,2718],{},"✅ MCP \u002F 工具调用刚需",[53,2720,2721],{},"❌ 单人桌面体验（用 Cherry Studio）",[53,2723,2724],{},"❌ 零运维 \u002F 不愿碰 Docker",[53,2726,2727],{},"❌ iOS 原生 App 主力",[30,2729,437],{"id":437},[50,2731,2732,2736,2740,2744],{},[53,2733,2734],{},[443,2735,943],{"href":24},[53,2737,2738],{},[443,2739,455],{"href":21},[53,2741,2742],{},[443,2743,952],{"href":526},[53,2745,2746],{},[443,2747,461],{"href":460},[30,2749,464],{"id":464},[197,2751,2752,2759,2766],{},[53,2753,2754,2755],{},"Open WebUI 官方文档 ",[443,2756,2757],{"href":2757,"rel":2758},"https:\u002F\u002Fdocs.openwebui.com\u002F",[474],[53,2760,2761,2762],{},"Local AI Master — Open WebUI Setup Guide 2026 ",[443,2763,2764],{"href":2764,"rel":2765},"https:\u002F\u002Flocalaimaster.com\u002Fblog\u002Fopen-webui-setup-guide",[474],[53,2767,2768,2769],{},"AIToolDiscovery — Set Up Open-WebUI with Ollama 2026 ",[443,2770,2771],{"href":2771,"rel":2772},"https:\u002F\u002Fwww.aitooldiscovery.com\u002Fhow-to\u002Fsetup-open-webui-ollama",[474],{"title":490,"searchDepth":491,"depth":491,"links":2774},[2775,2776,2777,2778,2779,2780,2781,2782,2783,2784],{"id":32,"depth":494,"text":33},{"id":48,"depth":494,"text":48},{"id":124,"depth":494,"text":124},{"id":2451,"depth":494,"text":2452},{"id":195,"depth":494,"text":195},{"id":233,"depth":494,"text":233},{"id":375,"depth":494,"text":375},{"id":410,"depth":494,"text":411},{"id":437,"depth":494,"text":437},{"id":464,"depth":494,"text":464},"\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。",[2788,2791,2794,2797],{"q":2789,"a":2790},"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":2792,"a":2793},"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":2795,"a":2796},"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":2798,"a":2799},"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。",[523,1010],{},[532,531,530,529,2803],"kubernetes",[2805,2807],{"plan":535,"price":536,"features":2806,"notes":1979},"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":656,"price":2808,"features":2809,"notes":2810},"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",[2813,2814],{"name":543,"url":460},{"name":545,"url":546},{"power":549,"ux":548,"price":549,"cn_support":548,"stability":549},{"title":755,"description":2786},"coding\u002Flocal\u002Fopen-webui",[2819,2821,2823],{"name":2820,"url":2757,"accessed":555},"Open WebUI 官方文档",{"name":2822,"url":2764,"accessed":555},"Local AI Master — Open WebUI Setup Guide 2026",{"name":2824,"url":2771,"accessed":555},"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",[504,2828,532,1044,2829,585,570],"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",[],1782316490685]