[{"data":1,"prerenderedAt":1029},["ShallowReactive",2],{"header-counts":3,"footer-counts":6,"compare-cherry-studio-vs-ollama":9,"compare-a-cherry-studio":10,"compare-b-ollama":539},{"tools":4,"reviews":5},78,26,{"tools":4,"reviews":5,"playbooks":7,"news":8},22,21,null,{"id":11,"title":12,"alternatives":13,"api_compatible":9,"body":18,"category":470,"chinese_friendly":471,"cover":472,"description":473,"domestic":474,"extension":475,"faq":476,"free":474,"github":9,"languages":489,"lastVerified":9,"meta":492,"models":9,"navigation":493,"notSuitable":9,"opensource":493,"path":494,"pillar":495,"platforms":496,"priceTable":501,"pricing":510,"published":511,"relatedPlaybooks":512,"relatedReviews":9,"score":515,"self_host":493,"seo":517,"seoTitle":9,"slug":518,"sources":519,"stem":527,"suitable":9,"tagline":528,"tags":529,"updated":522,"verdict":536,"website":537,"__hash__":538},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Fcherry-studio.md","Cherry Studio",[14,15,16,17],"coding\u002Flocal\u002Flobe-chat","coding\u002Flocal\u002Flm-studio","coding\u002Flocal\u002Follama","coding\u002Flocal\u002Fopen-webui",{"type":19,"value":20,"toc":455},"minimark",[21,26,30,33,36,83,86,100,106,110,115,132,137,154,157,184,187,338,341,373,377,400,403,429,432],[22,23,25],"h2",{"id":24},"tldr","TL;DR",[27,28,29],"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。",[27,31,32],{},"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。",[22,34,35],{"id":35},"核心能力",[37,38,39,47,53,59,65,71,77],"ul",{},[40,41,42,46],"li",{},[43,44,45],"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",[40,48,49,52],{},[43,50,51],{},"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",[40,54,55,58],{},[43,56,57],{},"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",[40,60,61,64],{},[43,62,63],{},"MCP 协议","：扩展工具调用 \u002F 联网搜索 \u002F 文件操作",[40,66,67,70],{},[43,68,69],{},"数据本地优先","：对话历史本地存储，WebDAV 同步，不上传第三方",[40,72,73,76],{},[43,74,75],{},"多模态","：图片识别 \u002F PDF 阅读 \u002F Markdown + Mermaid + 代码高亮",[40,78,79,82],{},[43,80,81],{},"AI 绘画 + 翻译","：内置主流 SD \u002F DALL·E \u002F 翻译 API 集成",[22,84,85],{"id":85},"价格",[37,87,88,94],{},[40,89,90,93],{},[43,91,92],{},"开源版","：完全免费，AGPL-3.0",[40,95,96,99],{},[43,97,98],{},"Enterprise","：私有化部署 + 团队协作 + 资源管控，联系销售",[101,102,103],"blockquote",{},[27,104,105],{},"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。",[22,107,109],{"id":108},"实测mac-m2-中型知识库","实测（Mac M2 + 中型知识库）",[27,111,112],{},[43,113,114],{},"亮点：",[37,116,117,120,123,126,129],{},[40,118,119],{},"中文 UI \u002F 文档 \u002F 社区都顶级，零门槛上手",[40,121,122],{},"本地 RAG 拖入 30+ PDF 后向量化 \u003C 2 分钟（用 bge-m3）",[40,124,125],{},"多模型并排回答：让 Claude \u002F GPT \u002F DeepSeek 同回一个问题做比较",[40,127,128],{},"MCP 接 Brave Search + 自定义工具流畅",[40,130,131],{},"WebDAV 同步坚果云 \u002F 阿里云盘，桌面 + 移动设备数据互通",[27,133,134],{},[43,135,136],{},"踩坑：",[37,138,139,142,145,148,151],{},[40,140,141],{},"没有 Web 端 \u002F Docker 自托管（要这个用 LobeChat）",[40,143,144],{},"iOS 版尚未发布（roadmap 中）",[40,146,147],{},"大型 PDF（>100 MB）向量化偶有失败，要切小",[40,149,150],{},"助手市场质量参差，要自筛",[40,152,153],{},"模型 API 调用全靠你自己付费，新手要先理解 API Key 概念",[22,155,156],{"id":156},"上手",[158,159,160,163,166,175,178,181],"ol",{},[40,161,162],{},"cherry-ai.com 下载客户端（或 GitHub releases）",[40,164,165],{},"设置 → 模型服务 → 填 OpenAI \u002F Claude \u002F DeepSeek API Key",[40,167,168,169],{},"（可选）本地：装 Ollama → Cherry Studio 自动识别 endpoint ",[170,171,172],"a",{"href":172,"rel":173},"http:\u002F\u002Flocalhost:11434",[174],"nofollow",[40,176,177],{},"新建知识库 → 拖文件 \u002F 加网址 → 等向量化",[40,179,180],{},"新对话 → 选模型 → 勾知识库 → 提问",[40,182,183],{},"进阶：自定义助手（System Prompt）+ MCP 扩展工具",[22,185,186],{"id":186},"对比",[188,189,190,211],"table",{},[191,192,193],"thead",{},[194,195,196,200,202,205,208],"tr",{},[197,198,199],"th",{},"维度",[197,201,12],{},[197,203,204],{},"LobeChat",[197,206,207],{},"LM Studio",[197,209,210],{},"Open WebUI",[212,213,214,231,245,261,274,290,305,321],"tbody",{},[194,215,216,220,223,226,228],{},[217,218,219],"td",{},"形态",[217,221,222],{},"桌面",[217,224,225],{},"Web + 桌面",[217,227,222],{},[217,229,230],{},"Docker \u002F 桌面",[194,232,233,235,238,240,243],{},[217,234,45],{},[217,236,237],{},"✅ 云 + 本地",[217,239,237],{},[217,241,242],{},"本地为主",[217,244,237],{},[194,246,247,250,253,255,258],{},[217,248,249],{},"知识库 RAG",[217,251,252],{},"✅ 强",[217,254,252],{},[217,256,257],{},"弱",[217,259,260],{},"✅",[194,262,263,266,268,270,272],{},[217,264,265],{},"MCP",[217,267,260],{},[217,269,260],{},[217,271,257],{},[217,273,260],{},[194,275,276,279,282,285,288],{},[217,277,278],{},"自托管 \u002F Web",[217,280,281],{},"无 Web",[217,283,284],{},"✅ Docker",[217,286,287],{},"无",[217,289,284],{},[194,291,292,295,298,300,303],{},[217,293,294],{},"中文",[217,296,297],{},"5\u002F5",[217,299,297],{},[217,301,302],{},"4\u002F5",[217,304,302],{},[194,306,307,310,313,316,319],{},[217,308,309],{},"开源协议",[217,311,312],{},"AGPL-3.0",[217,314,315],{},"MIT",[217,317,318],{},"闭源（免费）",[217,320,315],{},[194,322,323,326,329,332,335],{},[217,324,325],{},"GitHub Stars",[217,327,328],{},"60k+",[217,330,331],{},"72k+",[217,333,334],{},"–",[217,336,337],{},"126k+",[22,339,340],{"id":340},"避坑",[37,342,343,349,355,361,367],{},[40,344,345,348],{},[43,346,347],{},"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",[40,350,351,354],{},[43,352,353],{},"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",[40,356,357,360],{},[43,358,359],{},"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",[40,362,363,366],{},[43,364,365],{},"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",[40,368,369,372],{},[43,370,371],{},"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",[22,374,376],{"id":375},"适合-不适合","适合 \u002F 不适合",[37,378,379,382,385,388,391,394,397],{},[40,380,381],{},"✅ 中文用户、AI 重度使用 \u002F 多模型管理",[40,383,384],{},"✅ 需要本地 RAG 知识库",[40,386,387],{},"✅ 关注数据隐私 \u002F 本地存储",[40,389,390],{},"✅ 想用 Ollama \u002F LM Studio 本地模型",[40,392,393],{},"❌ 需要 Web 端 \u002F Docker 自托管",[40,395,396],{},"❌ 团队多人共享 \u002F SSO",[40,398,399],{},"❌ iOS 主力用户",[22,401,402],{"id":402},"相关阅读",[37,404,405,411,417,423],{},[40,406,407],{},[170,408,410],{"href":409},"\u002Ftools\u002Fcoding\u002Flocal\u002Flobe-chat","LobeChat 评测",[40,412,413],{},[170,414,416],{"href":415},"\u002Ftools\u002Fcoding\u002Flocal\u002Flm-studio","LM Studio 评测",[40,418,419],{},[170,420,422],{"href":421},"\u002Ftools\u002Fcoding\u002Flocal\u002Follama","Ollama 评测",[40,424,425],{},[170,426,428],{"href":427},"\u002Fplaybook\u002Fonboarding\u002Frag-pipeline-build","RAG Pipeline 搭建 Playbook",[22,430,431],{"id":431},"来源",[158,433,434,441,448],{},[40,435,436,437],{},"Cherry Studio 官网（功能 + 下载）",[170,438,439],{"href":439,"rel":440},"https:\u002F\u002Fwww.cherry-ai.com\u002F",[174],[40,442,443,444],{},"MBLUO Studio — Cherry Studio 评测 2026 ",[170,445,446],{"href":446,"rel":447},"https:\u002F\u002Fmbluostudio.com\u002Ftools\u002Fcherry-studio",[174],[40,449,450,451],{},"Cursor IDE 博客 — Cherry Studio 完全指南（2025-03）",[170,452,453],{"href":453,"rel":454},"https:\u002F\u002Fwww.cursor-ide.com\u002Fblog\u002Fcherry-studio-guide",[174],{"title":456,"searchDepth":457,"depth":457,"links":458},"",3,[459,461,462,463,464,465,466,467,468,469],{"id":24,"depth":460,"text":25},2,{"id":35,"depth":460,"text":35},{"id":85,"depth":460,"text":85},{"id":108,"depth":460,"text":109},{"id":156,"depth":460,"text":156},{"id":186,"depth":460,"text":186},{"id":340,"depth":460,"text":340},{"id":375,"depth":460,"text":376},{"id":402,"depth":460,"text":402},{"id":431,"depth":460,"text":431},"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",[477,480,483,486],{"q":478,"a":479},"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":481,"a":482},"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":484,"a":485},"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":487,"a":488},"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。",[490,491],"zh","en",{},true,"\u002Ftools\u002Fcoding\u002Flocal\u002Fcherry-studio","coding",[497,498,499,500],"windows","macos","linux","android",[502,506],{"plan":92,"price":503,"features":504,"notes":505},"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":98,"price":507,"features":508,"notes":509},"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",[513,514],"onboarding\u002Frag-pipeline-build","onboarding\u002Fcursor-mcp-deep-integration",{"power":516,"ux":471,"price":471,"cn_support":471,"stability":516},4,{"title":12,"description":473},"coding\u002Flocal\u002Fcherry-studio",[520,523,525],{"name":521,"url":439,"accessed":522},"Cherry Studio 官网","2026-06-24",{"name":524,"url":446,"accessed":522},"MBLUO Studio — Cherry Studio 评测",{"name":526,"url":453,"accessed":522},"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",[470,530,531,532,533,534,535],"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","LAkkQWZMoqlr7XC7II0hPyN5bwRJkF6rbuSjkfZz2eA",{"id":540,"title":541,"alternatives":542,"api_compatible":9,"body":543,"category":470,"chinese_friendly":457,"cover":981,"description":982,"domestic":474,"extension":475,"faq":983,"free":474,"github":9,"languages":996,"lastVerified":9,"meta":997,"models":9,"navigation":493,"notSuitable":9,"opensource":493,"path":421,"pillar":495,"platforms":998,"priceTable":1000,"pricing":1004,"published":511,"relatedPlaybooks":1005,"relatedReviews":9,"score":1007,"self_host":493,"seo":1008,"seoTitle":9,"slug":16,"sources":1009,"stem":1016,"suitable":9,"tagline":1017,"tags":1018,"updated":522,"verdict":1026,"website":1027,"__hash__":1028},"tools\u002Ftools\u002Fcoding\u002Flocal\u002Follama.md","Ollama",[15,17,518,14],{"type":19,"value":544,"toc":969},[545,547,555,558,560,630,632,635,639,643,663,667,698,700,734,736,862,864,896,898,921,923,944,946],[22,546,25],{"id":24},[27,548,549,550,554],{},"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，",[551,552,553],"code",{},"ollama pull"," 一键拉。",[27,556,557],{},"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）。",[22,559,35],{"id":35},[37,561,562,568,577,583,591,606,612,618,624],{},[40,563,564,567],{},[43,565,566],{},"后台 Daemon","：开机自启，应用调用零延迟",[40,569,570,573,574],{},[43,571,572],{},"CLI","：",[551,575,576],{},"ollama pull \u002F run \u002F list \u002F show \u002F create \u002F serve",[40,578,579,582],{},[43,580,581],{},"Modelfile","：类 Dockerfile 注册任意 GGUF，配 SYSTEM \u002F PARAMETER \u002F TEMPLATE",[40,584,585,573,588],{},[43,586,587],{},"OpenAI 兼容 API",[551,589,590],{},"http:\u002F\u002Flocalhost:11434\u002Fv1\u002Fchat\u002Fcompletions",[40,592,593,573,596,599,600,599,603],{},[43,594,595],{},"原生 API",[551,597,598],{},"\u002Fapi\u002Fchat","、",[551,601,602],{},"\u002Fapi\u002Fgenerate",[551,604,605],{},"\u002Fapi\u002Fembeddings",[40,607,608,611],{},[43,609,610],{},"模型库","：官方注册表内置 Llama \u002F Qwen \u002F DeepSeek \u002F Gemma \u002F Mistral \u002F GPT-OSS 等",[40,613,614,617],{},[43,615,616],{},"MLX 加速（Mac）","：0.19+ 起 M 系列自动用 MLX",[40,619,620,623],{},[43,621,622],{},"量化","：默认 Q4_K_M、支持 Q5 \u002F Q8 \u002F FP16",[40,625,626,629],{},[43,627,628],{},"跨平台","：Win \u002F Mac \u002F Linux 安装包，Docker 官方镜像",[22,631,85],{"id":85},[27,633,634],{},"完全免费、MIT 开源、商用免费。",[22,636,638],{"id":637},"实测m2-pro-qwen3-coder-7b-q4","实测（M2 Pro + Qwen3-Coder-7B Q4）",[27,640,641],{},[43,642,114],{},[37,644,645,651,654,657,660],{},[40,646,647,650],{},[551,648,649],{},"ollama run qwen3-coder:7b"," 一行起飞，3 秒进交互",[40,652,653],{},"REST API 配 Cursor \u002F Cline \u002F Continue 几乎全工具开箱即用",[40,655,656],{},"Modelfile 写自定义编码助手（low temperature + system prompt + 16K context）几分钟搞定",[40,658,659],{},"多模型并存，按需切换，内存占用合理",[40,661,662],{},"Mac M 系列 MLX 后比旧 GGUF 模式快显著",[27,664,665],{},[43,666,136],{},[37,668,669,679,685,692,695],{},[40,670,671,672,675,676],{},"默认 ",[551,673,674],{},"num_ctx"," 偏小（2048），跑长上下文要在 Modelfile 加 ",[551,677,678],{},"PARAMETER num_ctx 16384",[40,680,681,682],{},"模型默认走 0.0.0.0:11434 ↔ Docker 容器互访要 ",[551,683,684],{},"--add-host=host.docker.internal:host-gateway",[40,686,687,688,691],{},"国内 ",[551,689,690],{},"ollama.com\u002Flibrary"," 下载偶有慢，可手动 HF 下 GGUF + Modelfile 自建",[40,693,694],{},"多用户并发吞吐显著低于 vLLM",[40,696,697],{},"没有 GUI，模型浏览 \u002F 参数面板要走 LM Studio \u002F Open WebUI 配合",[22,699,156],{"id":156},[158,701,702,708,714,719,725,731],{},[40,703,704,707],{},[551,705,706],{},"curl -fsSL https:\u002F\u002Follama.ai\u002Finstall.sh | sh","（Mac \u002F Linux）；Windows winget",[40,709,710,713],{},[551,711,712],{},"ollama pull qwen3-coder:7b","（按需换模型）",[40,715,716,718],{},[551,717,649],{}," 直接聊",[40,720,721,722],{},"应用接入：baseURL = ",[551,723,724],{},"http:\u002F\u002Flocalhost:11434\u002Fv1",[40,726,727,728],{},"自定义：写 Modelfile → ",[551,729,730],{},"ollama create my-coder -f Modelfile",[40,732,733],{},"进阶：装 Open WebUI 做前端 \u002F 多人共享",[22,735,186],{"id":186},[188,737,738,754],{},[191,739,740],{},[194,741,742,744,746,748,751],{},[197,743,199],{},[197,745,541],{},[197,747,207],{},[197,749,750],{},"vLLM",[197,752,753],{},"llama.cpp",[212,755,756,772,787,801,816,832,847],{},[194,757,758,760,763,766,769],{},[217,759,219],{},[217,761,762],{},"CLI + Daemon",[217,764,765],{},"GUI + Headless",[217,767,768],{},"Python Server",[217,770,771],{},"C++ 二进制",[194,773,774,776,779,781,784],{},[217,775,156],{},[217,777,778],{},"极低",[217,780,778],{},[217,782,783],{},"中",[217,785,786],{},"高",[194,788,789,792,794,797,799],{},[217,790,791],{},"模型浏览",[217,793,572],{},[217,795,796],{},"✅ GUI",[217,798,287],{},[217,800,287],{},[194,802,803,806,809,812,814],{},[217,804,805],{},"OpenAI 兼容",[217,807,808],{},"✅ :11434",[217,810,811],{},"✅ :1234",[217,813,260],{},[217,815,260],{},[194,817,818,821,824,827,830],{},[217,819,820],{},"多用户吞吐",[217,822,823],{},"弱（~40 tok\u002Fs）",[217,825,826],{},"中（50–90）",[217,828,829],{},"强（800–12500）",[217,831,783],{},[194,833,834,837,840,842,845],{},[217,835,836],{},"MLX (Mac)",[217,838,839],{},"✅ 0.19+",[217,841,260],{},[217,843,844],{},"部分",[217,846,334],{},[194,848,849,852,854,857,860],{},[217,850,851],{},"开源",[217,853,315],{},[217,855,856],{},"闭源",[217,858,859],{},"Apache 2.0",[217,861,315],{},[22,863,340],{"id":340},[37,865,866,872,878,884,890],{},[40,867,868,871],{},[43,869,870],{},"num_ctx 一定要设","：默认 2K 太小，跑代码 \u002F 长文档要 16K+",[40,873,874,877],{},[43,875,876],{},"Modelfile 模板别漏 TEMPLATE","：错的 chat template 会让模型输出乱码 \u002F 不停",[40,879,880,883],{},[43,881,882],{},"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×",[40,885,886,889],{},[43,887,888],{},"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",[40,891,892,895],{},[43,893,894],{},"Mac 让它自动用 MLX","：升 0.19+；不要手动强制 GGUF + Metal",[22,897,376],{"id":375},[37,899,900,903,906,909,912,915,918],{},[40,901,902],{},"✅ 应用 \u002F IDE 接本地模型（Cursor \u002F Cline \u002F Continue）",[40,904,905],{},"✅ 个人 \u002F 评估 \u002F 脚本自动化",[40,907,908],{},"✅ Modelfile 自定义系统 prompt + 参数",[40,910,911],{},"✅ Mac M 系列 MLX 用户",[40,913,914],{},"❌ 多用户并发生产服务（用 vLLM）",[40,916,917],{},"❌ GUI 调参 \u002F 模型浏览（配 LM Studio \u002F Open WebUI）",[40,919,920],{},"❌ 极致单卡吞吐研究（直接 llama.cpp \u002F vLLM）",[22,922,402],{"id":402},[37,924,925,929,935,940],{},[40,926,927],{},[170,928,416],{"href":415},[40,930,931],{},[170,932,934],{"href":933},"\u002Ftools\u002Fcoding\u002Flocal\u002Fopen-webui","Open WebUI 评测",[40,936,937],{},[170,938,939],{"href":494},"Cherry Studio 评测",[40,941,942],{},[170,943,428],{"href":427},[22,945,431],{"id":431},[158,947,948,955,962],{},[40,949,950,951],{},"Markaicode — Import GGUF Models into Ollama 2026（2026-05-15）",[170,952,953],{"href":953,"rel":954},"https:\u002F\u002Fmarkaicode.com\u002Fimport-gguf-models-ollama-guide",[174],[40,956,957,958],{},"ComputingForGeeks — Ollama Models Cheat Sheet 2026 ",[170,959,960],{"href":960,"rel":961},"https:\u002F\u002Fcomputingforgeeks.com\u002Follama-models-cheat-sheet",[174],[40,963,964,965],{},"Codersera — Ollama vs LM Studio vs vLLM vs llama.cpp vs MLX 2026 ",[170,966,967],{"href":967,"rel":968},"https:\u002F\u002Fcodersera.com\u002Fblog\u002Follama-vs-lm-studio-vs-vllm-vs-llama-cpp-vs-mlx-2026\u002F",[174],{"title":456,"searchDepth":457,"depth":457,"links":970},[971,972,973,974,975,976,977,978,979,980],{"id":24,"depth":460,"text":25},{"id":35,"depth":460,"text":35},{"id":85,"depth":460,"text":85},{"id":637,"depth":460,"text":638},{"id":156,"depth":460,"text":156},{"id":186,"depth":460,"text":186},{"id":340,"depth":460,"text":340},{"id":375,"depth":460,"text":376},{"id":402,"depth":460,"text":402},{"id":431,"depth":460,"text":431},"\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。",[984,987,990,993],{"q":985,"a":986},"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":988,"a":989},"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":991,"a":992},"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":994,"a":995},"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。",[491],{},[497,498,499,999],"docker",[1001],{"plan":92,"price":503,"features":1002,"notes":1003},"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）",[513,1006],"onboarding\u002Fclaude-code-getting-started",{"power":516,"ux":516,"price":471,"cn_support":457,"stability":471},{"title":541,"description":982},[1010,1012,1014],{"name":1011,"url":953,"accessed":522},"Markaicode — Import GGUF 2026",{"name":1013,"url":960,"accessed":522},"ComputingForGeeks — Ollama Cheat Sheet 2026",{"name":1015,"url":967,"accessed":522},"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",[470,1019,1020,1021,1022,1023,1024,1025,534],"daemon","cli","rest-api","modelfile","gguf","mlx","openai-compatible","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","shirZzL900qiCQXzrJS1r3bv7XatM1q8hPc2mEoq88Y",1784565442360]