{
  "schemaVersion": "1.0",
  "updates": [],
  "companyNews": [],
  "id": "b636",
  "language": "zh",
  "name": "llama.cpp",
  "originalName": "llama.cpp",
  "library": "tools",
  "category": "模型开发",
  "categoryKey": "模型开发",
  "categoryKeys": [
    "模型开发"
  ],
  "categories": [
    "模型开发"
  ],
  "entryPoints": [],
  "summary": "以 C/C++ 运行本地模型，提供命令行、量化工具及带网页界面的模型服务器。",
  "introduction": "llama.cpp 是 ggml-org 维护的模型推理项目，不是名为 Llama 的单一模型。用户准备兼容的模型文件后，可通过命令行运行对话，或启动带网页界面和兼容 API 的服务器。项目支持 GGUF 模型工作流、量化和多种 CPU／GPU 后端，适合在本地或自有环境集成推理。模型文件、许可与硬件支持应分别核对，不能仅凭项目名称推断任意模型都能运行。",
  "capabilities": [],
  "company": {
    "name": "ggml-org",
    "originalName": "ggml-org"
  },
  "officialUrl": "https://github.com/ggml-org/llama.cpp",
  "profileUpdatedOn": "2026-10-22",
  "sources": [
    {
      "kind": "product",
      "url": "https://github.com/ggml-org/llama.cpp"
    },
    {
      "kind": "documentation",
      "url": "https://github.com/ggml-org/llama.cpp/tree/master/tools/server"
    }
  ],
  "news": [],
  "links": {
    "interactive": "products/llama-cpp--b636.html",
    "reading": "read/zh/products/llama-cpp--b636.html",
    "markdown": "read/zh/products/llama-cpp--b636.md",
    "json": "read/zh/products/llama-cpp--b636.json"
  }
}