LiteLLM Howto
LiteLLM is an open-source library and proxy server that provides a unified API for interacting with 100+ large language models from various providers using the OpenAI format.
View MoreHow to Use LiteLLM
Install LiteLLM: Install the LiteLLM library using pip: pip install litellm
Import and set up environment variables: Import litellm and set up environment variables for API keys: import litellm, os; os.environ['OPENAI_API_KEY'] = 'your-api-key'
Make an API call: Use the completion() function to make an API call: response = litellm.completion(model='gpt-3.5-turbo', messages=[{'role': 'user', 'content': 'Hello'}])
Handle streaming responses: For streaming responses, set stream=True: response = litellm.completion(model='gpt-3.5-turbo', messages=[{'role': 'user', 'content': 'Hello'}], stream=True)
Set up error handling: Use try-except blocks with OpenAIError to handle exceptions: try: litellm.completion(...) except OpenAIError as e: print(e)
Configure callbacks: Set up callbacks for logging: litellm.success_callback = ['helicone', 'langfuse']
Deploy LiteLLM Proxy: To deploy the LiteLLM proxy server, use Docker: docker run -e LITELLM_MASTER_KEY='sk-1234' ghcr.io/berriai/litellm:main
Configure model routing: Create a config.yaml file to set up model routing and API keys for different providers
Use the proxy server: Make API calls to your deployed LiteLLM proxy using the OpenAI SDK or curl commands
LiteLLM FAQs
LiteLLM is a unified API and proxy server that allows developers to interact with over 100 different LLM providers (like OpenAI, Azure, Anthropic, etc.) using a standardized OpenAI-compatible format. It simplifies LLM integration by providing features like load balancing, spend tracking, and consistent error handling across providers.
View More