The AI community's home for models, datasets, and Spaces
Hugging Face is the AI community's home for open-source models, datasets, and Spaces. With over 800K models, 200K datasets, and 300K Spaces, it's the largest repository of AI/ML resources. The Transformers library is the industry standard for working with pre-trained models. Hugging Face also offers Inference Endpoints for production deployment, AutoTrain for no-code fine-tuning, and the Hub API for programmatic access to the entire ecosystem.
800K+ pre-trained models: LLMs, vision, audio, multimodal. Download, fine-tune, and deploy with one line of code
Industry-standard Python library for loading and using pre-trained models. Supports PyTorch, TensorFlow, JAX, and ONNX
200K+ datasets for training and evaluation. Streaming, lazy loading, and efficient processing with the datasets library
300K+ hosted ML demos and apps. Build with Gradio or Streamlit, deploy instantly with free GPU/TPU for community projects
$0/mo
Best for open-source projects
$9/mo
For individual developers
Deploy any model as a production API on dedicated infrastructure. Auto-scaling, GPU selection, private networking
No-code model fine-tuning. Upload your data, select a base model, and train a custom model — no ML expertise required
High-performance LLM serving with continuous batching, tensor parallelism, and quantization. Powers the Inference API
Community benchmark tracking performance of open-source LLMs across standardized evaluations
Hugging Face Spaces now fully supports Gradio 4, enabling developers to build more advanced and user-friendly AI demos with new UI and accessibility features.
Fine-tune LLMs with no-code AutoTrain or LoRA/PEFT — from beginner to production.
From $20/user/mo
For organizations