RailRouter 🚂

A local LLM switchyard for macOS. See every model on your Mac, and send each request to the right engine through one OpenAI-compatible endpoint.

Ollama and LM Studio models in one table: size, quantization, capabilities. Click to copy an id.

What is loaded right now, RAM use, activity and prompt progress when the engine exposes it.

Search Ollama, Hugging Face and LM Studio catalogs. Filter by capability, sort by popularity or newest.

One address, 127.0.0.1:41235/v1. Routes by model to Ollama, LM Studio, MLX and free OpenRouter models.

Quick install

curl -fsSL https://raw.githubusercontent.com/Hugo291/RailRouter/main/install.sh | bash

Clones, builds and copies RailRouter to /Applications. Needs the Swift toolchain (xcode-select --install).

make test
make install
open /Applications/RailRouter.app

curl http://127.0.0.1:41235/v1/models

Requires macOS 14+ and at least one of Ollama, LM Studio or an MLX server. The endpoint listens on loopback only.