Amux provides 7 official adapters supporting mainstream LLM providers. Each adapter handles bidirectional conversion between provider-specific formats and a unified intermediate representation (IR).

Official Adapters

OpenAI

GPT-4, GPT-3.5, function calling, vision

Anthropic

Claude 3.5 Sonnet, Claude 3 Opus, 200K context

DeepSeek

DeepSeek Chat, DeepSeek Coder, high value

Moonshot

Kimi models, 200K ultra-long context

Zhipu

GLM-4 series, web search support

Qwen

Qwen series, vision and multimodal

Google Gemini

Gemini Pro, 1M+ tokens context

Feature Comparison

Adapter	Streaming	Tools	Vision	Multimodal	Reasoning	JSON Mode
OpenAI	✅	✅	✅	✅	❌	✅
Anthropic	✅	✅	✅	✅	✅	❌
DeepSeek	✅	✅	❌	❌	✅	✅
Moonshot	✅	✅	❌	❌	✅	✅
Zhipu	✅	✅	✅	✅	❌	✅
Qwen	✅	✅	✅	✅	✅	✅
Gemini	✅	✅	✅	✅	❌	✅

Choosing an Adapter

By Feature

Vision & Multimodal

OpenAI (GPT-4o), Anthropic (Claude 3), Qwen (Qwen-VL), Gemini (Gemini Pro Vision), Zhipu (GLM-4V)

Long Context

Gemini (1M+ tokens), Anthropic (200K), Moonshot (200K)

Coding

DeepSeek (DeepSeek Coder), OpenAI (GPT-4), Anthropic (Claude 3.5 Sonnet)

Reasoning

OpenAI (o3, o4-mini), DeepSeek (DeepSeek Reasoner), Moonshot (Kimi K2 Thinking)

By Price

Most Economical: DeepSeek, Qwen

Mid-range: OpenAI GPT-3.5, Moonshot

Premium: OpenAI GPT-4, Anthropic Claude, Gemini Pro

By OpenAI Compatibility

These adapters are OpenAI-compatible and can be seamlessly swapped:

DeepSeek (fully compatible)
Moonshot (fully compatible)
Zhipu (fully compatible)
Qwen (mostly compatible, minor differences)
Gemini (requires format conversion)

Quick Start

pnpm add @amux.ai/llm-bridge @amux.ai/adapter-openai @amux.ai/adapter-anthropic

Basic Usage

import { createBridge } from '@amux.ai/llm-bridge'
import { openaiAdapter } from '@amux.ai/adapter-openai'
import { anthropicAdapter } from '@amux.ai/adapter-anthropic'

// Accept OpenAI format, call Claude API
const bridge = createBridge({
  inbound: openaiAdapter,
  outbound: anthropicAdapter,
  config: {
    apiKey: process.env.ANTHROPIC_API_KEY
  }
})

const response = await bridge.chat({
  model: 'gpt-4',
  messages: [{ role: 'user', content: 'Hello!' }]
})

Creating Custom Adapters

Need to support other LLM providers? Create a custom adapter:

import type { LLMAdapter } from '@amux.ai/llm-bridge'

export const myAdapter: LLMAdapter = {
  name: 'my-provider',
  version: '1.0.0',
  capabilities: {
    streaming: true,
    tools: true,
    vision: false,
    multimodal: false,
    systemPrompt: true,
    toolChoice: true
  },
  inbound: {
    parseRequest: (request) => { /* ... */ },
    parseResponse: (response) => { /* ... */ },
    parseStream: (stream) => { /* ... */ },
    parseError: (error) => { /* ... */ }
  },
  outbound: {
    buildRequest: (ir) => { /* ... */ },
    buildResponse: (ir) => { /* ... */ }
  },
  getInfo() {
    return {
      name: this.name,
      version: this.version,
      capabilities: this.capabilities,
      endpoint: {
        baseURL: 'https://api.my-provider.com',
        chatPath: '/v1/chat'
      }
    }
  }
}

See the Custom Adapter Guide for a complete tutorial.

Next Steps

OpenAI Adapter

Learn how to use the OpenAI adapter

Custom Adapter Guide

Create your own adapter

Adapter API

View the complete Adapter API reference

Adapters Overview