Documentation

Getting Started

Absolute Router is a smart LLM proxy that automatically routes requests between fast and slow models based on prompt complexity. It's designed to be easy to use and integrate with existing OpenAI-compatible APIs.

Installation

You can use Absolute Router directly with your existing OpenAI client libraries. There's no need to install any additional packages.

Usage

There are two ways to configure the models for Absolute Router:

Using a single "model" parameter with a string format
Using separate "fastModel" and "slowModel" parameters with detailed configuration

1. Single Model Parameter

Single Model Parameter Example

import OpenAI from 'openai';

const openai = new OpenAI({
  apiKey: 'your-api-key',
  baseURL: 'https://router.abso.ai/v1',
});

const response = await openai.chat.completions.create({
  model: 'fast:gpt-4o|slow:o1-mini',  // This defines both fast and slow models
  messages: [{ role: 'user', content: 'What is the meaning of life?' }],
  stream: true,
});

for await (const chunk of response) {
  process.stdout.write(chunk.choices[0]?.delta?.content || '');
}

2. Separate Model Parameters

Separate Model Parameters Example

import OpenAI from 'openai';

const openai = new OpenAI({
  apiKey: 'your-api-key',
  baseURL: 'https://router.abso.ai/v1',
});

const response = await openai.chat.completions.create({
  model: 'gpt-4o',  // This can be any model name, it will be overridden
  messages: [{ role: 'user', content: 'What is the meaning of life?' }],
  stream: true,
  metadata: {
    fastModel: {
      name: "gpt-4o",
      apiUrl: "https://api.openai.com/v1",
      apiKey: "your-openai-api-key"
    },
    slowModel: {
      name: "o1-mini",
      apiUrl: "https://api.anthropic.com/v1",
      apiKey: "your-anthropic-api-key"
    }
  }
});

for await (const chunk of response) {
  process.stdout.write(chunk.choices[0]?.delta?.content || '');
}

API Reference

Absolute Router is compatible with the OpenAI API. For detailed information on available parameters and response formats, please refer to the OpenAI API documentation.