[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"package:cask:ollama-binary:zh":3,"releases:stats:cask:ollama-binary:zh":310,"releases:cask:ollama-binary:1:false:false:zh":311,"releases-count:cask:ollama-binary:true:zh":395},{"artifacts":4,"autoUpdates":25,"categories":26,"conflictsWith":32,"dependsOn":36,"deprecated":25,"description":47,"descriptionEn":48,"developer":49,"disabled":25,"displayName":50,"downloadSha256":51,"downloadSize":52,"downloadUrl":53,"editorChoice":25,"formulaeUrl":54,"homepage":55,"iconUrl":56,"installCommand":57,"installs":58,"installs30d":59,"isFont":25,"isLibrary":25,"kegOnly":25,"kind":60,"latestRelease":61,"machineTranslated":63,"minMacos":46,"name":49,"names":69,"platforms":71,"primaryCategory":283,"rank30d":284,"releaseCount":285,"releaseStats":286,"repoUrl":292,"screenshots":293,"sourceLocale":294,"sourceUrl":295,"summary":296,"summaryTranslation":297,"supports":299,"tags":301,"tap":308,"token":70,"version":68,"versionChangedAt":309},{"apps":5,"binaries":6,"entries":8,"pkgs":24},[],[7],"ollama",[9,16],{"declaration":10,"phase":13,"sources":14,"target":12,"type":15},{"binary":11,"target":12},[7],"$HOMEBREW_PREFIX\u002Fbin\u002Follama","install",[7],"binary",{"declaration":17,"phase":21,"sources":22,"type":23},{"zap":18},[19],{"trash":20},"~\u002F.ollama","cleanup",[],"zap",[],false,[27],{"icon":28,"machineTranslated":25,"name":29,"slug":30,"sourceLocale":31},"lucide:sparkles","AI 工具","ai","zh-CN",{"casks":33,"formulae":35},[34],"ollama-app",[],{"arch":37,"casks":40,"formulae":41,"macos":42,"requirements":43},[38,39],"arm64","x86_64",[],[],">= 14",{"macos":44},{">=":45},[46],"14","Ollama 提供用于运行开放模型的命令行工作流，以及供需要模型推理的应用使用的 REST API。这个特定的 Homebrew 软件包 ollama-binary 安装的是 ollama 可执行文件，而非 Ollama.app。它与 macOS 上独立的 ollama-app cask 冲突；桌面应用的截图不应被视为此 CLI 发行版的截图。\n\n## 模型工作流\n使用命令行入口选择模型，或连接受支持的编程智能体集成。官方 README 介绍了如何交互式运行模型、启动 Claude Code 或 Codex 等集成，以及在 Ollama 模型库中查找模型。模型下载与运行时归档文件相互独立，因此软件包的下载大小并不代表模型集合所需的存储空间。\n\n## API 与集成\n本地 REST API 支持运行和管理模型，文档中的聊天示例使用 localhost:11434\u002Fapi\u002Fchat。官方 Python 和 JavaScript 库提供可编程调用的客户端。社区界面和开发工具可以连接到 Ollama，但它们是独立产品，并非此 cask 捆绑提供的图形界面。\n\n## Apple 芯片运行时与要求\n版本 0.40.0 会在 Apple 芯片设备上自动为该运行时支持的架构使用 MLX，包括文档中列出的聊天、决策和嵌入模型。版本 0.40.1 移除了 CLI 初始设置中的账户步骤，并为云端用量和余额 API 添加了服务器代理功能。涉及云服务的功能与本地运行模型不同。Homebrew 当前记录的版本为 0.40.1，分发官方 Darwin 归档文件，并要求 macOS Sonoma 14 或更高版本。\n\n## 来源\n- [官方仓库与 README](https:\u002F\u002Fgithub.com\u002Follama\u002Follama)\n- [CLI 参考文档](https:\u002F\u002Fdocs.ollama.com\u002Fcli)\n- [API 参考文档](https:\u002F\u002Fdocs.ollama.com\u002Fapi)\n- [官方 0.40.0 版本说明](https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Freleases\u002Ftag\u002Fv0.40.0)\n- [官方 0.40.1 版本说明](https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Freleases\u002Ftag\u002Fv0.40.1)\n- [Homebrew 二进制 cask](https:\u002F\u002Fgithub.com\u002FHomebrew\u002Fhomebrew-cask\u002Fblob\u002FHEAD\u002FCasks\u002Fo\u002Follama-binary.rb)\n","Get up and running with large language models locally","Ollama","Ollama CLI","e888b7637291ceb80b622c00ea067f62c86d9c50d419b3eb903032e4f1f8a4f6",167473220,"https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Freleases\u002Fdownload\u002Fv0.40.2\u002Follama-darwin.tgz","https:\u002F\u002Fformulae.brew.sh\u002Fcask\u002Follama-binary","https:\u002F\u002Follama.com\u002F","https:\u002F\u002Fcdn.opennavo.com\u002Ficons\u002Fuploads\u002F8a0bba297e0c\u002Fb3f619860729-256.png","brew install --cask ollama-binary",{"d30":59,"d365":59,"d90":59},207,"cask",{"brewCommittedAt":62,"hasNotes":25,"isLatest":63,"isPrerelease":25,"sections":64,"source":65,"translation":66,"version":68},"2026-10-09T04:38:34Z",true,[],"homebrew",{"status":67},"none","0.40.2",[49,70],"ollama-binary",[72,99,125,152,178,205,231,257],{"arch":39,"artifacts":73,"conflictsWith":87,"dependsOn":90,"downloadSha256":51,"downloadUrl":53,"macos":97,"minMacos":46,"requiresRosetta":25,"tag":98,"version":68},{"apps":74,"binaries":75,"entries":76,"pkgs":86},[],[7],[77,81],{"declaration":78,"phase":13,"sources":80,"target":12,"type":15},{"binary":79,"target":12},[7],[7],{"declaration":82,"phase":21,"sources":85,"type":23},{"zap":83},[84],{"trash":20},[],[],{"casks":88,"formulae":89},[34],[],{"arch":91,"casks":92,"formulae":93,"macos":42,"requirements":94},[39],[],[],{"macos":95},{">=":96},[46],"27","golden_gate",{"arch":38,"artifacts":100,"conflictsWith":114,"dependsOn":117,"downloadSha256":51,"downloadUrl":53,"macos":97,"minMacos":46,"requiresRosetta":25,"tag":124,"version":68},{"apps":101,"binaries":102,"entries":103,"pkgs":113},[],[7],[104,108],{"declaration":105,"phase":13,"sources":107,"target":12,"type":15},{"binary":106,"target":12},[7],[7],{"declaration":109,"phase":21,"sources":112,"type":23},{"zap":110},[111],{"trash":20},[],[],{"casks":115,"formulae":116},[34],[],{"arch":118,"casks":119,"formulae":120,"macos":42,"requirements":121},[38],[],[],{"macos":122},{">=":123},[46],"arm64_golden_gate",{"arch":39,"artifacts":126,"conflictsWith":140,"dependsOn":143,"downloadSha256":51,"downloadUrl":53,"macos":150,"minMacos":46,"requiresRosetta":25,"tag":151,"version":68},{"apps":127,"binaries":128,"entries":129,"pkgs":139},[],[7],[130,134],{"declaration":131,"phase":13,"sources":133,"target":12,"type":15},{"binary":132,"target":12},[7],[7],{"declaration":135,"phase":21,"sources":138,"type":23},{"zap":136},[137],{"trash":20},[],[],{"casks":141,"formulae":142},[34],[],{"arch":144,"casks":145,"formulae":146,"macos":42,"requirements":147},[39],[],[],{"macos":148},{">=":149},[46],"26","tahoe",{"arch":38,"artifacts":153,"conflictsWith":167,"dependsOn":170,"downloadSha256":51,"downloadUrl":53,"macos":150,"minMacos":46,"requiresRosetta":25,"tag":177,"version":68},{"apps":154,"binaries":155,"entries":156,"pkgs":166},[],[7],[157,161],{"declaration":158,"phase":13,"sources":160,"target":12,"type":15},{"binary":159,"target":12},[7],[7],{"declaration":162,"phase":21,"sources":165,"type":23},{"zap":163},[164],{"trash":20},[],[],{"casks":168,"formulae":169},[34],[],{"arch":171,"casks":172,"formulae":173,"macos":42,"requirements":174},[38],[],[],{"macos":175},{">=":176},[46],"arm64_tahoe",{"arch":39,"artifacts":179,"conflictsWith":193,"dependsOn":196,"downloadSha256":51,"downloadUrl":53,"macos":203,"minMacos":46,"requiresRosetta":25,"tag":204,"version":68},{"apps":180,"binaries":181,"entries":182,"pkgs":192},[],[7],[183,187],{"declaration":184,"phase":13,"sources":186,"target":12,"type":15},{"binary":185,"target":12},[7],[7],{"declaration":188,"phase":21,"sources":191,"type":23},{"zap":189},[190],{"trash":20},[],[],{"casks":194,"formulae":195},[34],[],{"arch":197,"casks":198,"formulae":199,"macos":42,"requirements":200},[39],[],[],{"macos":201},{">=":202},[46],"15","sequoia",{"arch":38,"artifacts":206,"conflictsWith":220,"dependsOn":223,"downloadSha256":51,"downloadUrl":53,"macos":203,"minMacos":46,"requiresRosetta":25,"tag":230,"version":68},{"apps":207,"binaries":208,"entries":209,"pkgs":219},[],[7],[210,214],{"declaration":211,"phase":13,"sources":213,"target":12,"type":15},{"binary":212,"target":12},[7],[7],{"declaration":215,"phase":21,"sources":218,"type":23},{"zap":216},[217],{"trash":20},[],[],{"casks":221,"formulae":222},[34],[],{"arch":224,"casks":225,"formulae":226,"macos":42,"requirements":227},[38],[],[],{"macos":228},{">=":229},[46],"arm64_sequoia",{"arch":39,"artifacts":232,"conflictsWith":246,"dependsOn":249,"downloadSha256":51,"downloadUrl":53,"macos":46,"minMacos":46,"requiresRosetta":25,"tag":256,"version":68},{"apps":233,"binaries":234,"entries":235,"pkgs":245},[],[7],[236,240],{"declaration":237,"phase":13,"sources":239,"target":12,"type":15},{"binary":238,"target":12},[7],[7],{"declaration":241,"phase":21,"sources":244,"type":23},{"zap":242},[243],{"trash":20},[],[],{"casks":247,"formulae":248},[34],[],{"arch":250,"casks":251,"formulae":252,"macos":42,"requirements":253},[39],[],[],{"macos":254},{">=":255},[46],"sonoma",{"arch":38,"artifacts":258,"conflictsWith":272,"dependsOn":275,"downloadSha256":51,"downloadUrl":53,"macos":46,"minMacos":46,"requiresRosetta":25,"tag":282,"version":68},{"apps":259,"binaries":260,"entries":261,"pkgs":271},[],[7],[262,266],{"declaration":263,"phase":13,"sources":265,"target":12,"type":15},{"binary":264,"target":12},[7],[7],{"declaration":267,"phase":21,"sources":270,"type":23},{"zap":268},[269],{"trash":20},[],[],{"casks":273,"formulae":274},[34],[],{"arch":276,"casks":277,"formulae":278,"macos":42,"requirements":279},[38],[],[],{"macos":280},{">=":281},[46],"arm64_sonoma",{"icon":28,"machineTranslated":25,"name":29,"slug":30,"sourceLocale":31},527,9,{"brewLag":287,"cadence":291,"count30d":285},{"compared":288,"earlierCount":289,"medianMinutes":290},3,0,4325,"daily","https:\u002F\u002Fgithub.com\u002Follama\u002Follama",[],"en-US","https:\u002F\u002Fgithub.com\u002Fhomebrew\u002Fhomebrew-cask\u002Fblob\u002FHEAD\u002FCasks\u002Fo\u002Follama-binary.rb","Ollama 命令行发行版，用于运行开放模型、管理本地推理并提供 REST API；此 cask 不会安装桌面应用。",{"locale":31,"machineTranslated":63,"sourceLocale":294,"status":298},"machine",{"arm64":63,"requiresRosetta":25,"status":300,"x86_64":63},"known",[302,303,304,305,306,307],"Local AI","CLI","LLM","REST API","MLX","Apple Silicon","homebrew\u002Fcask","2026-10-09T07:15:26.193153Z",null,{"current":312,"records":313,"size":392,"stats":393,"total":285},1,[314,317,335,350,367,372,377,382,387],{"brewCommittedAt":62,"hasNotes":25,"isLatest":63,"isPrerelease":25,"sections":315,"source":65,"translation":316,"version":68},[],{"status":67},{"bodyMarkdown":318,"brewCommittedAt":319,"hasNotes":63,"id":320,"isLatest":25,"isPrerelease":25,"machineTranslated":63,"publishedAt":321,"sections":322,"source":330,"sourceLocale":294,"summary":331,"title":332,"translation":333,"version":334},"## 更新内容\r\n* server：代理云端用量和余额 API，由 @drifkin 提交，见 https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fpull\u002F18829\r\n* llama：修复 windows 上 clef 头部读取超过 2GiB 的问题，由 @Gigrise 提交，见 https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fpull\u002F18777\r\n* cmd：移除 CLI 初始设置中的账户步骤，由 @hoyyeva 提交，见 https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fpull\u002F18826\r\n* manifest：避免在 Windows 上使用符号链接，由 @dhiltgen 提交，见 https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fpull\u002F18852\r\n* docs：修复 README 社区集成列表中的 6 个失效链接，由 @aniketkrs 提交，见 https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fpull\u002F18814\r\n* docs：修复应用 README 中失效的下载链接，由 @chenlichao 提交，见 https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fpull\u002F18233\r\n* mlx：移除已被上游合并的 metal 驻留补丁，由 @dhiltgen 提交，见 https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fpull\u002F18854\r\n\r\n## 新贡献者\r\n* @Gigrise 首次贡献，见 https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fpull\u002F18777\r\n* @aniketkrs 首次贡献，见 https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fpull\u002F18814\r\n* @chenlichao 首次贡献，见 https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fpull\u002F18233\r\n\r\n**完整更新日志**：https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fcompare\u002Fv0.40.0...v0.40.1-rc0\n\n来源：https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Freleases\u002Ftag\u002Fv0.40.1","2026-10-08T05:37:31Z",1737,"2026-10-07T23:22:59Z",[323],{"area":324,"items":325},"变更",[326,327,328,329],"移除 CLI 初始设置中的账户步骤。","通过服务器代理云端用量和余额 API。","修复 Windows 上超出 2 GiB 位置的 Clef 头部读取问题。","避免在 Windows 上使用清单符号链接。","editorial","移除 CLI 初始设置中的账户步骤，代理云端用量和余额 API，并修复 Windows 上的 Clef 读取和清单符号链接问题。","Ollama 0.40.1",{"locale":31,"machineTranslated":63,"sourceLocale":294,"status":298},"0.40.1",{"bodyMarkdown":336,"brewCommittedAt":337,"hasNotes":63,"id":338,"isLatest":25,"isPrerelease":25,"machineTranslated":63,"publishedAt":339,"sections":340,"source":330,"sourceLocale":294,"summary":346,"title":347,"translation":348,"version":349},"## 更新内容\r\n\r\n**模型默认在 Apple Silicon 上使用 MLX 运行**\r\n\r\n在此版本中，Apple Silicon 设备上受 MLX 运行时支持的模型架构将自动使用 MLX 运行。\r\n\r\n```\r\nollama pull qwen3.8\r\nollama run qwen3.8\r\n```\r\n\r\n新增模型包括 [gemma4](https:\u002F\u002Follama.com\u002Flibrary\u002Fgemma4)、[qwen3.6](https:\u002F\u002Follama.com\u002Flibrary\u002Fqwen3.6) 和 [qwen3.5](https:\u002F\u002Follama.com\u002Flibrary\u002Fqwen3.5)\r\n\r\n决策模型现在也可在 MLX 上运行：[Nimble](https:\u002F\u002Follama.com\u002Flibrary\u002Fnimble) [tev1](https:\u002F\u002Follama.com\u002Flibrary\u002Ftev1)  [clef](https:\u002F\u002Follama.com\u002Flibrary\u002Fclef) [clef-flash](https:\u002F\u002Follama.com\u002Flibrary\u002Fclef-flash) \r\n\r\nMLX 现在支持嵌入模型：[embeddinggemma-2](https:\u002F\u002Follama.com\u002Flibrary\u002Fembeddinggemma-2)\r\n\r\n我们将继续测试并启用更多模型。\r\n\r\n\r\n**完整更新日志**：https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fcompare\u002Fv0.35.1...v0.40.0\n\n来源：https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Freleases\u002Ftag\u002Fv0.40.0","2026-10-06T19:26:20Z",1739,"2026-09-25T03:31:52Z",[341],{"area":324,"items":342},[343,344,345],"在 Apple Silicon 上默认使用 MLX 运行受支持的架构。","支持在 MLX 上运行决策模型。","新增在 MLX 上对 embeddinggemma-2 的支持。","在 Apple Silicon 上通过 MLX 自动运行受支持的模型架构，包括聊天和决策模型，并新增一个 MLX 嵌入模型。","Ollama 0.40.0",{"locale":31,"machineTranslated":63,"sourceLocale":294,"status":298},"0.40.0",{"bodyMarkdown":351,"brewCommittedAt":352,"hasNotes":63,"id":353,"isLatest":25,"isPrerelease":25,"machineTranslated":63,"publishedAt":354,"sections":355,"source":330,"sourceLocale":294,"summary":363,"title":364,"translation":365,"version":366},"## Clef 决策模型\r\n\r\nOllama 现已通过 `\u002Fv1\u002Fsystemone` 支持 Cloudflare 新推出的开源决策模型 [Clef](https:\u002F\u002Follama.com\u002Flibrary\u002Fclef) 和 [Clef Flash](https:\u002F\u002Follama.com\u002Flibrary\u002Fclef)。\r\n\r\nClef (27B) 和 Clef Flash (9B) 均为多模态模型：请求现在可以在文本状态之外包含图像，所有问题共享这些图像，并将其与文本状态一起评分。\r\n\r\n```sh\r\ncurl http:\u002F\u002Flocalhost:11434\u002Fv1\u002Fsystemone -d '{\r\n  \"model\": \"clef-flash\",\r\n  \"state\": \"The user took this screenshot.\",\r\n  \"images\": [\"\u003Cbase64-encoded image>\"],\r\n  \"questions\": {\r\n    \"has_ollama\": {\"type\": \"noul\", \"instructions\": \"Does this image contain Ollama?\"}\r\n  }\r\n}'\r\n```\r\n\r\n```\r\n{\r\n  \"model\": \"clef-flash\",\r\n  \"answers\": {\r\n    \"has_ollama\": {\r\n      \"type\": \"noul\",\r\n      \"noul\": 0.958\r\n    }\r\n  },\r\n  \"usage\": {\r\n    \"input_tokens\": 548,\r\n    \"output_tokens\": 0\r\n  }\r\n}\r\n```\r\n## 更新内容\r\n* 使用网页搜索的模型现在每次回复最多可执行十次搜索，此前为三次\r\n* Modelfile 现在支持 `CAPABILITY` 声明，让模型创建者可以明确声明模型的能力。从 GGUF 或 safetensors 创建模型、继承模型以及导出 Modelfile 时，都会保留这些声明\r\n* `ollama show` 和模型列表现在仅将 `decision` 报告为决策模型的能力，因此客户端不再将其用于通用聊天、工具调用或思考\r\n* 更新了 llama.cpp 和 MLX 引擎\r\n\r\n**完整更新日志**：https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Fcompare\u002Fv0.35.0...v0.35.1\r\n\n\n来源：https:\u002F\u002Fgithub.com\u002Follama\u002Follama\u002Freleases\u002Ftag\u002Fv0.35.1","2026-10-02T20:20:16Z",1741,"2026-09-29T20:14:22Z",[356],{"area":324,"items":357},[358,359,360,361,362],"通过 \u002Fv1\u002Fsystemone 添加多模态 Clef 和 Clef Flash。","将每次响应的网页搜索次数从三次增加到十次。","保留 Modelfile 中的显式 CAPABILITY 声明。","为决策模型报告仅用于决策的能力。","更新 llama.cpp 和 MLX 引擎。","添加多模态 Clef 决策模型，将每次响应的网页搜索次数增加到十次，并保留 Modelfile 能力声明；更新推理引擎。","Ollama 0.35.1",{"locale":31,"machineTranslated":63,"sourceLocale":294,"status":298},"0.35.1",{"brewCommittedAt":368,"hasNotes":25,"isLatest":25,"isPrerelease":25,"sections":369,"source":65,"translation":370,"version":371},"2026-09-30T03:09:20Z",[],{"status":67},"0.35.0",{"brewCommittedAt":373,"hasNotes":25,"isLatest":25,"isPrerelease":25,"sections":374,"source":65,"translation":375,"version":376},"2026-09-24T07:51:18Z",[],{"status":67},"0.34.4",{"brewCommittedAt":378,"hasNotes":25,"isLatest":25,"isPrerelease":25,"sections":379,"source":65,"translation":380,"version":381},"2026-09-22T22:28:58Z",[],{"status":67},"0.34.3",{"brewCommittedAt":383,"hasNotes":25,"isLatest":25,"isPrerelease":25,"sections":384,"source":65,"translation":385,"version":386},"2026-09-18T02:12:14Z",[],{"status":67},"0.34.2",{"brewCommittedAt":388,"hasNotes":25,"isLatest":25,"isPrerelease":25,"sections":389,"source":65,"translation":390,"version":391},"2026-09-15T22:23:43Z",[],{"status":67},"0.34.1",20,{"brewLag":394,"cadence":291,"count30d":285},{"compared":288,"earlierCount":289,"medianMinutes":290},{"current":312,"records":396,"size":312,"stats":400,"total":285},[397],{"brewCommittedAt":62,"hasNotes":25,"isLatest":63,"isPrerelease":25,"sections":398,"source":65,"translation":399,"version":68},[],{"status":67},{"brewLag":401,"cadence":291,"count30d":285},{"compared":288,"earlierCount":289,"medianMinutes":290}]