Hugging Face has introduced a new Messages API that enables developers to interact with open large language models (LLMs) using the same interface as OpenAI's API. This move aims to simplify the transition from proprietary models like GPT to open alternatives, which can be hosted on Hugging Face's Inference Endpoints or other platforms.
The Messages API supports chat-based interactions, allowing users to send a sequence of messages with system, user, and assistant roles. It works with popular open LLMs such as Llama, Mistral, and Falcon, providing a drop-in replacement for OpenAI's chat completions endpoint.
Developers can switch between models without changing code by simply updating the endpoint URL and API key. Hugging Face has also published detailed documentation and a migration guide to help users adapt their existing applications.
This initiative addresses growing demand for open-source AI models that offer transparency, customization, and lower costs compared to closed proprietary systems. By standardizing the API format, Hugging Face aims to accelerate adoption of open LLMs across industries.