LLMrest

A lightweight, unified JavaScript library for interacting with multiple LLM providers including OpenAI, Anthropic Claude, Google Gemini, and Perplexity.

Features

• Unified API - Single interface to work with multiple LLM providers
• Support for major models - Works with GPT models, Claude models, Gemini models, and Llama models
• Middleware support - Built-in content moderation and input validation
• Error handling - Robust error handling with detailed error messages
• Concurrent requests - Make parallel requests to multiple models
• Flexible messaging - Send different messages to different models

Installation

npm install @armagank/llmrest

Quick Start

const { createAIClient } = require('@armagank/llmrest');

// Initialize with your API keys
const client = createAIClient({
  apiKeys: {
    openai: 'your-openai-api-key',
    claude: 'your-anthropic-api-key',
    gemini: 'your-google-api-key',
    perplexity: 'your-perplexity-api-key'
  }
});

// Use the same messages for multiple models
async function chatWithModels() {
  const messages = [
    { role: 'system', content: 'You are a helpful assistant.' },
    { role: 'user', content: 'What is the capital of France?' }
  ];
  
  const response = await client.createChat({
    models: ['gpt-4', 'claude-3-opus-20240229'],
    messages,
    maxInput: 100000,  // bytes
    maxOutput: 1000,   // tokens
    moderationEnabled: true
  });
  
  console.log(response);
}

// Use different messages for different models
async function chatWithCustomMessages() {
  const modelMessages = {
    'gpt-4': [
      { role: 'system', content: 'You are a helpful AI assistant.' },
      { role: 'user', content: 'Explain quantum computing.' }
    ],
    'claude-3-opus-20240229': [
      { role: 'system', content: 'You are Claude, an AI assistant by Anthropic.' },
      { role: 'user', content: 'Explain quantum computing in simple terms.' }
    ]
  };
  
  const response = await client.createChatMessages({
    models: ['gpt-4', 'claude-3-opus-20240229'],
    messages: modelMessages,
    maxInput: 100000,
    maxOutput: 1000,
    moderationEnabled: false
  });
  
  console.log(response);
}

API Reference

createAIClient(options)

Creates a new LLMrest client.

Parameter	Type	Description
options.apiKeys	Object	Object containing API keys for each provider

Returns an object with the following methods:

client.createChat(options)

Send the same messages to multiple models.

Parameter	Type	Description
options.models	Array<string>	Array of model identifiers
options.messages	Array<Object>	Array of message objects
options.maxInput	number	Maximum input size in bytes
options.maxOutput	number	Maximum output tokens
options.moderationEnabled	boolean	Whether to run content moderation

client.createChatMessages(options)

Send different messages to different models.

Parameter	Type	Description
options.models	Array<string>	Array of model identifiers
options.messages	Object	Object mapping model IDs to message arrays
options.maxInput	number	Maximum input size in bytes
options.maxOutput	number	Maximum output tokens
options.moderationEnabled	boolean	Whether to run content moderation

Supported Models

• OpenAI: Models starting with gpt- (e.g., gpt-4, gpt-3.5-turbo)
• Anthropic: Models starting with claude- (e.g., claude-3-opus-20240229, claude-3-sonnet-20240229)
• Google: Models starting with gemini- (e.g., gemini-pro, gemini-ultra)
• Perplexity: Models starting with llama- (e.g., llama-3-70b-instruct)

Error Handling

LLMrest handles errors gracefully and provides detailed error messages. The response will include error details for any model that failed to process.

// Example response with an error
{
  'gpt-4': 'The capital of France is Paris.',
  'claude-3-opus-20240229': { error: 'API key is invalid.' }
}

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.