omnilabs 0.0.1

A Python client for the OmniRouter API

OmniLabs Python Client

A powerful Python client for the OmniRouter API that provides unified access to various LLM models through a single interface. This client supports chat completions, image generation, and smart model selection capabilities.

Installation

pip install omnilabs

Quick Start

from omnilabs import OmniClient, ChatMessage

# Initialize the client
client = OmniClient()  # API key can be set via OMNI_API_KEY environment variable
# Or provide the API key directly
client = OmniClient(api_key="your-api-key")

# Simple chat completion
messages = [
    ChatMessage(role="user", content="What is the capital of France?")
]
response = client.chat(messages, model="gpt-4")

# Generate an image
image_response = client.generate_image(
    prompt="A serene landscape with mountains at sunset",
    model="dall-e-3",
    size="1024x1024"
)

# Smart model selection
smart_response = client.smart_select(
    messages=messages,
    k=5,  # Get top 5 recommended models
    rel_cost=0.5,
    rel_accuracy=0.5
)

Features

1. Chat Completions

• Support for various chat models (e.g., GPT-4, Claude, etc.)
• Customizable parameters like temperature and max tokens
• Streaming support for real-time responses

2. Image Generation

• Support for models like DALL-E 3
• Multiple size options: 256x256, 512x512, 1024x1024
• Quality settings: standard and HD

3. Smart Model Selection

• Intelligent model recommendations based on your requirements
• Balance between cost, latency, and accuracy
• Customizable weights for different factors

API Reference

ChatMessage Class

ChatMessage(role: str, content: str)

Represents a single message in a chat conversation.

OmniClient Class

OmniClient(api_key: Optional[str] = None, base_url: str = "https://omni-router.vercel.app/")

Methods:

1. chat(messages, model, temperature=0.7, max_tokens=None, stream=False)

   • Send chat messages and get completions
   • Supports streaming for real-time responses

2. generate_image(prompt, model="dall-e-3", size="1024x1024", quality="standard", n=1)

   • Generate images from text prompts
   • Multiple size and quality options

3. smart_select(messages, k=5, model_names=None, rel_cost=0.5, rel_latency=0.0, rel_accuracy=0.5)

   • Get model recommendations based on your requirements
   • Customize importance of cost, latency, and accuracy

4. get_available_models(model_type=None)

   • List available models
   • Optional filtering by model type ('chat' or 'image')

Environment Variables

• OMNI_API_KEY: Your OmniLabs API key

Error Handling

The client includes proper error handling for common scenarios:

• Invalid API keys
• Network issues
• Rate limiting
• Invalid parameters

Contributing

We welcome contributions! Please check our GitHub repository for guidelines.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Support

• Documentation: https://omnilabs-ai.github.io
• Issues: GitHub Issues