llm-flow-router

Tiny Express-like multiplexer for routing AI prompts to multiple providers (Grok, OpenAI, Anthropic, Gemini) with middleware support.

Perfect for reliable hybrid AI apps, fallbacks, logging, and rate-limiting in 2026.

Features

• Route prompts like Express: /gemini, /grok, /openai, etc.
• Middleware chain: logger, fallback, custom ones
• Built-in providers with unified chat format
• Fallback routes for zero-downtime
• Lightweight: <10 files, minimal deps

Installation

npm install llm-flow-router

Badges

Author: Divesh Sarkar

Setup

Create .env in your project:

Set env vars: OPENAI_API_KEY, GROK_API_KEY, ANTHROPIC_API_KEY, GEMINI_API_KEY

Usage

import { mux } from 'llm-flow-router';
import 'dotenv/config';  // if not already loaded

const messages = [
  { role: 'user', content: 'Explain quantum computing like I\'m 12.' }
];

async function demo() {
  // Fast & cheap: Gemini 2.5 Flash
  const gemini = await mux.handle({ path: '/gemini', messages });
  console.log('Gemini:', gemini);

  // Reliable fallback
  const safe = await mux.handle({ path: '/safe', messages });
  console.log('Safe fallback:', safe);
}

demo();

Built-in Routes

/grok → Grok (xAI)
/openai → OpenAI
/claude → Anthropic Claude
/safe → Fallback: Gemini → Grok → OpenAI
/default → Gemini

Custom Middleware Example

mux.use(async (prompt, next) => {
  console.log('Custom middleware running');
  return next();
});

MIT License