Openrouter provider

[Johnson-f]

OpenRouter Provider Guide

OpenRouter is a unified API that provides access to 200+ AI models from multiple providers including OpenAI, Anthropic, Google, Meta, Microsoft, and more. The chimeric OpenRouter provider gives you seamless access to this vast ecosystem of models through a single, consistent interface.

🌟 Key Features

• 200+ Models: Access models from OpenAI, Anthropic, Google, Meta, Microsoft, Perplexity, and more
• Cost Optimization: Often cheaper than direct provider APIs with transparent pricing
• Unified Interface: Same API for all models, no need to learn different SDKs
• Model Fallbacks: Automatic failover if a model is unavailable
• Free Models: Access to free models for testing and development
• Higher Rate Limits: Better rate limits than individual provider APIs
• Streaming Support: Real-time response streaming for all compatible models
• Tool Calling: Function calling support for models that support it
• Async Support: Full async/await support for high-performance applications

📦 Installation

# Install chimeric with OpenAI support (OpenRouter is OpenAI-compatible)
pip install "chimeric[openai]"

# Or with uv (recommended)
uv add "chimeric[openai]"

# For development with all extras
uv add "chimeric[all]"

🔐 Authentication

Get Your API Key

1. Visit OpenRouter.ai
2. Sign up for a free account
3. Navigate to the API Keys section
4. Generate a new API key

Set Up Environment Variable

# Set environment variable (recommended)
export OPENROUTER_API_KEY="your-api-key-here"

# Or add to your .env file
echo "OPENROUTER_API_KEY=your-api-key-here" >> .env

🚀 Quick Start

Basic Usage

from chimeric import Chimeric

# Initialize with environment variable
client = Chimeric()

# Or pass API key directly
client = Chimeric(openrouter_api_key="your-api-key")

# Simple text generation
response = client.generate(
    model="openai/gpt-4o-mini",
    messages="Hello! Explain quantum computing in simple terms."
)

print(response.content)

List Available Models

# Get all available models
models = client.list_models("openrouter")
print(f"Total models available: {len(models)}")

# Show first few models
for model in models[:5]:
    print(f"- {model.id}: {model.name}")

🎯 Model Selection

OpenRouter provides access to models from many providers. Here are some popular choices:

OpenAI Models

# GPT-4o models (latest and most capable)
response = client.generate(
    model="openai/gpt-4o",
    messages="Write a Python function to calculate fibonacci numbers"
)

# GPT-4o-mini (faster, cheaper)
response = client.generate(
    model="openai/gpt-4o-mini", 
    messages="Summarize the benefits of renewable energy"
)

[Verdenroz]

I tried testing the tools + streaming but found that I was getting no chunks at all. Do you have the cassettes from the integration test or a code snippet of what model you used to test tools + streaming with your OpenRouterClient?

[Johnson-f]

Will look into it

[Verdenroz]

Closing this as I have refactored the entire library to flatten dependencies. I have included Openrouter in #12.