Quick answer: Model hub + datasets + fine-tuning + Inference API — the AI commons.
Hugging Face is an open-source platform and community hub built by Hugging Face Inc. that democratizes access to machine learning models, datasets, and tools. It serves as the central repository for pre-trained transformer models across NLP, computer vision, and audio tasks. The platform provides a model hub with thousands of community-contributed models, a datasets library for training data, built-in fine-tuning capabilities, and an Inference API for deploying models without managing infrastructure. Hugging Face operates on a freemium model, offering free hosting and API access for open-source projects while providing paid options for private models and higher usage. It's become the de facto standard for sharing and discovering ML models, with support for frameworks like PyTorch and TensorFlow, making it essential infrastructure for AI development.