🤖 Multi-AI Wrapper

The Universal API for Generative AI.

Use OpenAI (GPT-4), Anthropic (Claude 3.5), and Google Gemini through a single, unified interface. Never worry about API downtime again with built-in automatic fallbacks. Monitor every penny with real-time cost tracking.

Build reliable, cost-effective AI applications in minutes, not days.

🌟 GitHub Repository: https://github.com/Mehulbirare/multi-ai-wrapper

⚡ Why Use Multi-AI Wrapper?

Developers choose multi-ai-wrapper over LangChain or official SDKs because it solves the biggest problems in AI integration:

| Feature | multi-ai-wrapper | Official SDKs | LangChain | | :--- | :---: | :---: | :---: | | Unified API | ✅ Yes | ❌ No | ✅ Yes | | Automatic Fallback | ✅ Built-in | ❌ Manual | ⚠️ Complex | | Cost Tracking | ✅ Real-time (USD) | ❌ No | ❌ Difficult | | Lightweight | ✅ Extremely | ✅ Yes | ❌ Heavy | | Streaming | ✅ Unified Stream | ⚠️ Varies | ✅ Yes |

Key Benefits

1. Zero Downtime (High Availability): If OpenAI is down, your app automatically switches to Anthropic or Gemini. Your users never see an error.
2. Cost Optimization: Track spending per provider and model. Switch to cheaper models instantly with aliases like "cheap" or "smart".
3. No Vendor Lock-in: Write your code once. Switch providers by changing one line of configuration.
4. Developer Experience: Simple, intuitive API that just works. No complex abstractions or boilerplate.

📦 Installation

npm install multi-ai-wrapper

🚀 Quick Start

1. Initialize

Create a single instance for your entire app.

const AI = require('multi-ai-wrapper');

const ai = new AI({
  providers: [
    { name: 'openai', apiKey: process.env.OPENAI_API_KEY, priority: 1 },    // Primary
    { name: 'anthropic', apiKey: process.env.ANTHROPIC_API_KEY, priority: 2 } // Backup
  ]
});

2. Chat (Unified API)

const response = await ai.chat({
  message: "Explain quantum computing to a 5-year-old",
  model: "smart", // Automatically uses GPT-4o or Claude 3.5 Sonnet
  temperature: 0.7
});

console.log(response.text);
console.log(`Provider used: ${response.provider}`); // e.g., "openai"
console.log(`Cost: $${response.cost}`); // e.g., "$0.0024"

3. Stream (Real-time)

const stream = await ai.chatStream({
  message: "Write a futuristic story...",
  model: "fast" // Automatically uses GPT-4o-mini or Gemini Flash
});

for await (const chunk of stream) {
  process.stdout.write(chunk.text);
}

🔧 Advanced Features

Smart Model Aliases

Don't memorize model names. Use aliases that map to the best option for each provider:

• "smart": Best capability (GPT-4o, Claude 3.5 Sonnet, Gemini 1.5 Pro)
• "fast": Low latency (GPT-4o-mini, Claude Haiku, Gemini Flash)
• "cheap": Lowest cost (GPT-4o-mini, Gemini Flash)

Detailed Cost Analytics

const stats = ai.getCostStats();
console.log(stats);
/*
{
  totalCost: 0.145,
  requestCount: 52,
  byProvider: {
    openai: { cost: 0.120, requests: 40 },
    anthropic: { cost: 0.025, requests: 12 }
  }
}
*/

📚 Documentation

For a deep dive into architecture and advanced configuration, see the Detailed Explanation & Guide.

🤝 Contributing

We welcome contributions! Whether it's adding a new provider (Mistral, Cohere, etc.) or fixing a bug, please check out our repository.

📄 License

MIT © Mehulbirare