@ipconfiger/llm-hub
v0.1.0
Published
A local LLM proxy that routes requests to multiple backends by model name.
Maintainers
Readme
@ipconfiger/llm-hub
A local LLM proxy that routes requests to multiple backends by model name.
Install
npm install -g @ipconfiger/llm-hub
llm-hub --serveAt install time a postinstall script automatically downloads the correct
prebuilt binary for your platform from GitHub Releases.
Supported platforms
| OS | Architecture |
| ------- | ------------ |
| macOS | arm64 (Apple Silicon), x64 (Intel) |
| Linux | x64, arm64 (musl static default; set LLM_HUB_VARIANT=gnu for glibc) |
| Windows | x64 |
Environment variables
| Variable | Default | Description |
| -------- | ------- | ----------- |
| LLM_HUB_VARIANT | (unset) | Set to gnu on Linux to download the glibc-linked binary instead of musl. |
| LLM_HUB_FORCE_DOWNLOAD | (unset) | Set to 1 to force re-download even if the binary already exists. |
Requirements
- Node.js 18+ (only for install/download; the binary itself is standalone)
- If your platform has no prebuilt binary, build from source: LLM-HUB
Usage
llm-hub --serve # start the HTTP proxy server
llm-hub --admin # TUI admin interface
llm-hub --init # create a sample config
llm-hub --bind 0.0.0.0:8080 # custom bind address (with --serve)See the project README for full docs.
License
MIT
