Pi-AI: Unified LLM API

Integrate diverse Large Language Models (LLMs) effortlessly with Pi-AI's unified API. This Action Pack shows you how to install Pi-AI, configure providers with automatic model discovery, and seamlessly switch between models using a consistent API.

beginner5 min3 steps

The play

Install Pi-AI
Install the `pi-ai` Python package using pip. This provides the core library for unified LLM interaction.
Initialize LLMClient with Provider Configurations
Instantiate the `LLMClient`, providing your API keys for the desired LLM providers. Pi-AI automatically discovers available models and their optimal configurations.
Interact with LLMs and Switch Models
Use the `client.chat.completions.create` method to interact with any configured LLM. Easily switch models and providers by changing the `model` parameter, leveraging the consistent API structure.

Starter code

from pi_ai import LLMClient

# Replace with your actual API keys
OPENAI_API_KEY = "YOUR_OPENAI_KEY"
ANTHROPIC_API_KEY = "YOUR_ANTHROPIC_KEY"

client = LLMClient(
    provider_config={
        "openai": {"api_key": OPENAI_API_KEY},
        "anthropic": {"api_key": ANTHROPIC_API_KEY}
    }
)

# Example 1: Use an OpenAI model
response_gpt = client.chat.completions.create(
    model="gpt-4o",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "What is the capital of France?"}
    ]
)
print(f"OpenAI Response: {response_gpt.choices[0].message.content}")

# Example 2: Use an Anthropic model
response_claude = client.chat.completions.create(
    model="claude-3-opus-20240229",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Tell me a fun fact about Paris."}
    ]
)
print(f"Anthropic Response: {response_claude.choices[0].message.content}")