mcpdirectory.dev
9,129 servers · updated 23 Sept

Freellmpool MCP server

community Active last commit 4 d
{
  "mcpServers": {
    "0xzr-freellmpool": {
      "command": "uvx",
      "args": [
        "freellmpool"
      ],
      "env": {
        "OPENAI_BASE_URL": "sk-...",
        "ANTHROPIC_BASE_URL": "sk-ant-...",
        "GROQ_API_KEY": "<groq_api_key>",
        "FREELLMPOOL_PROXY_KEY": "<freellmpool_proxy_key>",
        "FREELLMPOOL_JOBS_PATH": "/path/to/directory",
        "LLM7_API_KEY": "<llm7_api_key>"
      }
    }
  }
}
file .mcp.json (in your project root)install uvx freellmpoolversion 0.13.0

What it does

freellmpool catalogs 22 LLM providers across free tiers, keyless endpoints, and trials, exposing 178 enabled chat routes and 431 models. It pools accessible routes behind an OpenAI-compatible endpoint available as a CLI, Python library, or local proxy. It works without credentials when keyless providers are available.

  • Ask questions and get replies from free LLM models via command line with automatic provider failover and rate-limit handling
  • Set up a local proxy endpoint that speaks OpenAI API to run coding agents and other tools against pooled free models
  • Discover and import local LM Studio, Ollama, or llama.cpp instances and route requests to them

Before you enable it

An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.