[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"pack-detail-self-hosted-ai-zh":3,"seo:pack:self-hosted-ai:zh":66},{"code":4,"message":5,"data":6},200,"操作成功",{"pack":7},{"slug":8,"icon":9,"tone":10,"status":11,"status_label":12,"title":13,"description":14,"items":15,"install_cmd":65},"self-hosted-ai","🏠","#059669","stable","稳定","本地部署 AI","Tabby \u002F Onyx \u002F LibreChat + n8n 启动器 — 数据全留自己机器上。",[16,28,36,44,51,58],{"id":17,"uuid":18,"slug":19,"title":20,"description":21,"author_name":22,"view_count":23,"vote_count":24,"lang_type":25,"type":26,"type_label":27},216,"1a1d4061-a148-4566-a3d7-ab40e6f2a972","tabby-self-hosted-ai-coding-assistant-1a1d4061","Tabby — Self-Hosted AI Coding Assistant","Self-hosted AI code completion and chat assistant. Privacy-first alternative to GitHub Copilot. Supports 20+ models, repo-aware context, and IDE integrations. 33K+ stars.","TokRepo精选",1141,0,"en","skill","Skill",{"id":29,"uuid":30,"slug":31,"title":32,"description":33,"author_name":34,"view_count":35,"vote_count":24,"lang_type":25,"type":26,"type_label":27},390,"e1fd7c46-bbda-4956-8649-9c3ed579ff25","whisper-cpp-local-speech-text-pure-c-c-e1fd7c46","whisper.cpp — Local Speech-to-Text in Pure C\u002FC++","High-performance port of OpenAI Whisper in C\u002FC++. No Python, no GPU required. Runs on CPU, Apple Silicon, CUDA, and even Raspberry Pi. Real-time transcription.","Script Depot",1949,{"id":37,"uuid":38,"slug":39,"title":40,"description":41,"author_name":42,"view_count":43,"vote_count":24,"lang_type":25,"type":26,"type_label":27},321,"210679a0-712f-4ec5-8d69-e0a016361c95","onyx-self-hosted-ai-chat-40-connectors-210679a0","Onyx — Self-Hosted AI Chat with 40+ Connectors","Onyx (formerly Danswer) is a self-hosted AI chat with RAG, custom agents, and 40+ knowledge connectors. 20.4K+ stars. Enterprise search. MIT.","AI Open Source",383,{"id":45,"uuid":46,"slug":47,"title":48,"description":49,"author_name":42,"view_count":50,"vote_count":24,"lang_type":25,"type":26,"type_label":27},284,"850494fb-7737-4388-8104-f8860a0d2d41","librechat-self-hosted-multi-ai-chat-platform-850494fb","LibreChat — Self-Hosted Multi-AI Chat Platform","LibreChat is a self-hosted AI chat platform unifying Claude, OpenAI, Google, AWS in one interface. 35.1K+ GitHub stars. Agents, MCP, code interpreter, multi-user auth. MIT.",318,{"id":52,"uuid":53,"slug":54,"title":55,"description":56,"author_name":42,"view_count":57,"vote_count":24,"lang_type":25,"type":26,"type_label":27},483,"92d3cc62-6199-4b1c-a7f1-1b73a1da86a0","self-hosted-ai-starter-kit-local-ai-n8n-92d3cc62","Self-Hosted AI Starter Kit — Local AI with n8n","Docker Compose template by n8n that bootstraps a complete local AI environment with n8n workflow automation, Ollama LLMs, Qdrant vector database, and PostgreSQL. 14,500+ stars.",369,{"id":59,"uuid":60,"slug":61,"title":62,"description":63,"author_name":42,"view_count":64,"vote_count":24,"lang_type":25,"type":26,"type_label":27},870,"f05a11a5-33e5-11f1-9bc6-00163e2b0d79","typebot-visual-ai-chatbot-builder-you-can-self-host-f05a11a5","Typebot — Visual AI Chatbot Builder You Can Self-Host","Build advanced chatbots visually with 34+ blocks. Embed anywhere, collect results in real-time. OpenAI integration, custom themes, analytics. Self-hostable. 9,800+ stars.",349,"tokrepo install pack\u002Fself-hosted-ai",{"pageType":67,"pageKey":8,"locale":68,"title":69,"metaDescription":70,"h1":13,"tldr":71,"bodyMarkdown":72,"faq":73,"schema":89,"internalLinks":99,"citations":112,"wordCount":125,"generatedAt":126},"pack","zh","本地部署 AI：Tabby \u002F Onyx \u002F LibreChat \u002F n8n 套装","Tabby \u002F Onyx \u002F LibreChat \u002F n8n —— 六个自托管 AI 资产替代 Copilot \u002F ChatGPT \u002F Zapier，数据全留自己机器。TokRepo 一条命令装齐。","六个生产级自托管 AI 资产 —— Tabby（Copilot 替代）\u002F Onyx（企业搜索）\u002F LibreChat（ChatGPT 团队版）+ n8n AI 启动器。数据全留自己机器。","## 这个 pack 装了什么\n\n这个 pack 收齐了**六个自托管 AI 资产**，是团队为合规、成本或主权原因离开 SaaS 时反复出现的选择。三个是编码 \u002F 聊天替代品（Tabby \u002F LibreChat \u002F Onyx），三个是基础设施块（n8n AI 启动器、本地 STT、模型网关）。\n\n| # | 资产 | 类型 | 替代什么 |\n|---|---|---|---|\n| 1 | Tabby | 自托管服务 | GitHub Copilot |\n| 2 | Onyx | 自托管服务 | Glean \u002F 企业版 ChatGPT |\n| 3 | LibreChat | 自托管 UI | 团队的 ChatGPT |\n| 4 | n8n AI 启动器 | docker-compose | 带 AI 节点的 Zapier |\n| 5 | Whisper STT（本地） | 服务 | Otter \u002F Rev \u002F 云 STT |\n| 6 | 本地模型网关 | 服务 | LiteLLM 本地优先路由 |\n\n## 为什么要装\n\n2026 年默认 AI 栈假设你愿意把代码、聊天、客户数据发给 OpenAI \u002F Anthropic \u002F Google。多数 C 端应用没问题。但医疗 \u002F 金融 \u002F 法律等受监管行业、政府工作，或者团队的 IP *就是*产品的，根本不能。这个 pack 给出组装好的答案：一个能在单工作站或小 K8s 集群跑的栈，给你 Copilot 级开发工具、ChatGPT 级聊天、企业搜索级检索 —— 全在自己硬件上。\n\n三个头部替代品：\n\n- **Tabby** 是 Copilot 的替代。自托管、IDE 指过来，你得到由本地模型（DeepSeek-Coder \u002F Qwen-Coder 等）支撑的内联补全。一张 3090 多数语言能匹配 Copilot 质量。\n- **Onyx**（前身 Danswer）是企业搜索的替代。接上你的 Confluence \u002F Notion \u002F GitHub \u002F Slack，搭一个内部 ChatGPT 从你的文档里答问。向量 + 关键词混合搜索带引用。\n- **LibreChat** 是团队 ChatGPT 替代。多用户、多模型（搭本地 Ollama 或云 API 兜底）、对话历史、prompt 库。要给团队一个「ChatGPT」又不想按席位付费，就用这个。\n\n三个基础设施块补缺口。n8n 启动器给你 n8n + Postgres + Qdrant + 本地模型的 docker-compose —— 自家机器上的工作流自动化。本地 Whisper 让会议转写和语音备忘永不离开你的网。模型网关在本地和云模型间路由，本地搞不定时再回退到 Claude。\n\n## 一条命令装齐\n\n```bash\n# 装整个 pack\ntokrepo install pack\u002Fself-hosted-ai\n\n# 或挑你真要的\ntokrepo install tabby\ntokrepo install onyx\ntokrepo install librechat\ntokrepo install n8n-ai-starter-kit\n```\n\nTokRepo CLI 装好 docker-compose 文件、环境模板，以及给你 AI 工具的规则 \u002F subagent —— 告诉它什么时候调本地栈、什么时候上云。装完 `docker compose up -d`，服务在 localhost 就通了。\n\n## 常见坑\n\n- **别用 16GB 显存跑 70B 模型**。模型大小对显卡。Tabby 的 DeepSeek-Coder-7B 12GB 卡能跑，补全足够。聊天用 4-bit Qwen-2.5-32B 在 24GB 是甜点。\n- **Onyx 连接器会静默限速**。把 Onyx 指向 5 万页 Confluence 时，首次同步几小时起步，部分连接器会暂停。盯日志；前 24 小时别信 UI 进度条。\n- **n8n + AI 工作流会泄露凭据**。启动器默认 Postgres 密码明文。改掉，并在暴露前把 n8n 放 Cloudflare Tunnel 或带认证的反代后面。\n- **LibreChat 默认权限扁平**。开箱每个用户能看每个对话。上团队前先配 RBAC 和按用户的模型白名单。\n- **备份不是自动的**。自托管 = 自备份。给 LibreChat \u002F Onyx 排 pg_dump，给 Tabby 模型缓存做快照；存储预算按活跃数据 3 倍算还原点。\n\n## 跟其他 pack 的关系\n\n这个 pack 跟两个 pack 天然搭配。**MCP 服务器全家桶** 给你协议层连接器（文件系统 \u002F 浏览器 \u002F 数据库 MCP server）路由经过你本地模型网关 —— 这样连 Claude Code 都能调你的本地服务。**LLM 可观测性** 这里比云 API 更重要，因为故障面是你自己的；那个 pack 里的 Langfuse 自托管版能干净接入 Onyx 和 LibreChat。\n\n从零起的话，装顺序：1) LibreChat（即时用户价值）；2) Tabby（开发者价值）；3) Onyx（组织级搜索）；4) 在上面建自动化时再装 n8n + 网关。",[74,77,80,83,86],{"q":75,"a":76},"Tabby 免费吗？","免费。Tabby 开源 Apache 2.0，自托管 Community 版免费。有付费 Enterprise 版含 SSO、审计日志、SLA，但 Community 版功能完整够个人和小团队用。你只付跑它的 GPU 钱。Onyx \u002F LibreChat \u002F n8n 都是同模型 —— 全开源，可选付费。",{"q":78,"a":79},"Cursor 或 Codex CLI 能用吗？","自托管服务工具无关 —— Tabby 暴露 Copilot 兼容 API，任何支持 Copilot 的 IDE 都能接（VS Code \u002F JetBrains \u002F Vim）。LibreChat 是 web UI 工具独立。TokRepo CLI 装对应工具的配置（Cursor 规则 \u002F AGENTS.md \u002F Claude Code subagent）告诉 agent 本地服务存在。",{"q":81,"a":82},"Tabby 跟 Cursor + 本地模型怎么选？","Cursor 本地模型支持限于特定端点；Tabby 是为自托管代码补全专门设计的，带遥测、模型预热、真后端。要 IDE 无关、多团队自托管 Copilot，Tabby 赢。要 Cursor 的 UX 后面接本地模型，看这个 pack 里的本地模型网关 —— 能扮 Cursor 兼容端点。",{"q":84,"a":85},"跟 MCP 服务器全家桶 pack 有啥区别？","MCP 全家桶讲协议层连接器，让 AI 工具能读你的文件系统 \u002F 浏览器 \u002F 数据库。本地部署 AI 是把云 LLM \u002F UI \u002F IDE 助手整体替换成自家硬件上的服务。两者互补：MCP server 可配置成路由经过你本地模型网关，给你完全自有 agent 栈。",{"q":87,"a":88},"什么时候*不*该自托管？","延迟比主权更重要时（实时语音、小模型 sub-300ms 代码补全难）；用量低到 GPU 划不来（每月 $100 API 调用比 4090 三年摊销便宜）；或没运维支持搞备份、模型升级、和注定的凌晨 2 点 OOM。自托管是真运维活，预算要算上。",{"@context":90,"@type":91,"name":92,"description":93,"numberOfItems":94,"publisher":95},"https:\u002F\u002Fschema.org","CollectionPage","Self-Hosted AI","Tabby, Onyx, LibreChat, and an n8n starter kit — keep your data on your own metal.",6,{"@type":96,"name":97,"url":98},"Organization","TokRepo","https:\u002F\u002Ftokrepo.com",[100,104,108],{"url":101,"anchor":102,"reason":103},"\u002Fzh\u002Fpacks\u002Fmcp-server-stack","MCP 服务器全家桶","MCP server 把本地模型接进 AI 工具",{"url":105,"anchor":106,"reason":107},"\u002Fzh\u002Fpacks\u002Fllm-observability","LLM 可观测性","监控你的自托管栈",{"url":109,"anchor":110,"reason":111},"\u002Fzh\u002Ftools\u002Fcline","Cline","VS Code agent 跟本地 Tabby 搭配良好",[113,117,121],{"claim":114,"source_name":115,"source_url":116},"Tabby is a self-hosted AI coding assistant alternative to GitHub Copilot","TabbyML\u002Ftabby on GitHub","https:\u002F\u002Fgithub.com\u002FTabbyML\u002Ftabby",{"claim":118,"source_name":119,"source_url":120},"LibreChat is an open-source ChatGPT clone supporting multiple LLM backends","danny-avila\u002FLibreChat on GitHub","https:\u002F\u002Fgithub.com\u002Fdanny-avila\u002FLibreChat",{"claim":122,"source_name":123,"source_url":124},"n8n is a fair-code workflow automation platform with self-hosting support","n8n.io\u002Fself-hosted","https:\u002F\u002Fdocs.n8n.io\u002Fhosting\u002F",560,"2026-05-02T15:00:00Z"]