What is Amux?

Amux is a bidirectional API adapter that enables seamless conversion between different LLM provider formats.

Key benefits:

Provider flexibility - Switch between OpenAI, Anthropic, DeepSeek, and others without changing your code
Bidirectional conversion - Accept requests in any format, call any provider, return responses in any format
Type-safe - Full TypeScript support with comprehensive type definitions
Zero dependencies - Core package has no runtime dependencies
Production-ready - Used in production applications

Unlike other solutions that only provide a unified interface, Amux enables true bidirectional conversion - you control both the input and output formats.

Quick Example

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

// Create a bridge: OpenAI format in → Anthropic API out
const bridge = createBridge({
  inbound: openaiAdapter,
  outbound: anthropicAdapter,
  config: {
    apiKey: process.env.ANTHROPIC_API_KEY
  }
})

// Send OpenAI-format request, get OpenAI-format response
// But actually calls Claude API under the hood
const response = await bridge.chat({
  model: 'gpt-4',
  messages: [{ role: 'user', content: 'Hello!' }]
})

console.log(response.choices[0].message.content)

How It Works

Amux uses an Intermediate Representation (IR) pattern to convert between any provider formats:

Your App (OpenAI format)
    ↓
Inbound Adapter → Parse to IR
    ↓
Intermediate Representation (unified format)
    ↓
Outbound Adapter → Build Anthropic format
    ↓
Claude API
    ↓
Response flows back through IR
    ↓
Your App (OpenAI format)

The IR is a unified format that captures all LLM capabilities - messages, tools, streaming, multimodal content, and provider-specific extensions.

Supported Providers

Provider	Package	Features
OpenAI	`@amux.ai/adapter-openai`	Chat, streaming, tools, vision, JSON mode
Anthropic	`@amux.ai/adapter-anthropic`	Messages, streaming, tools, vision, reasoning
DeepSeek	`@amux.ai/adapter-deepseek`	Chat, streaming, tools, reasoning
Moonshot	`@amux.ai/adapter-moonshot`	Chat, streaming, tools
Zhipu (GLM)	`@amux.ai/adapter-zhipu`	Chat, streaming, tools
Qwen	`@amux.ai/adapter-qwen`	Chat, streaming, tools, vision
Gemini	`@amux.ai/adapter-google`	Chat, streaming, tools, vision

Common Use Cases

Multi-provider support - Let users choose their preferred LLM provider:

const providers = {
  openai: createBridge({ inbound: openaiAdapter, outbound: openaiAdapter }),
  anthropic: createBridge({ inbound: openaiAdapter, outbound: anthropicAdapter })
}

const response = await providers[userChoice].chat(request)

Provider migration - Switch providers without code changes:

// Old: Using OpenAI
const bridge = createBridge({
  inbound: openaiAdapter,
  outbound: openaiAdapter
})

// New: Switch to Anthropic (no other code changes needed)
const bridge = createBridge({
  inbound: openaiAdapter,
  outbound: anthropicAdapter
})

Cost optimization - Route to cheaper providers for simple requests:

const isSimple = request.messages.length < 3
const outbound = isSimple ? deepseekAdapter : anthropicAdapter

const bridge = createBridge({ inbound: openaiAdapter, outbound })