Hugging Face

The GitHub of AI — 700,000+ models, datasets, and Spaces to explore and deploy

Hugging Face is the central hub of the open-source AI world. It hosts over 700,000 models (text, vision, audio, code), 150,000+ datasets, and interactive demos called Spaces. Developers use it to find pre-trained models, run them via API, fine-tune them on custom data, and share their own work.

🧠

Models

700,000+

Pre-trained models for text, images, audio, video, and code

📊

Datasets

150,000+

Training and benchmark datasets for all AI tasks

🚀

Spaces

300,000+

Live interactive demos — try any model in your browser

💻

Inference API

Pay-per-use

Call any model via REST API without managing infrastructure

⚡ Getting Started

Create a free account

Go to huggingface.co → Sign Up. The free tier gives you access to the model hub, Spaces, and limited Inference API usage.

Browse Models

huggingface.co/models — filter by task (text generation, translation, image classification etc.), language, and license. Sort by "Most Downloads" to find popular ones.

Try a model in Spaces

Many models have a "Spaces" demo button — click it to test the model live in your browser with no code. Great for evaluation before committing.

Get your API token

Settings → Access Tokens → New Token. Use this token to call models via the Inference API or download models programmatically.

🚀 Sample Usage

Call the Inference API (Python)

pip install huggingface_hub

from huggingface_hub import InferenceClient

client = InferenceClient(
  model="mistralai/Mistral-7B-Instruct-v0.1",
  token="hf_YOUR_TOKEN_HERE"
)

response = client.text_generation(
  "Explain recursion in simple terms",
  max_new_tokens=200
)
print(response)

Download and Run a Model Locally

pip install transformers torch

from transformers import pipeline

# Load a text classification model
classifier = pipeline("sentiment-analysis")
result = classifier("This product is absolutely amazing!")
# Output: [{'label': 'POSITIVE', 'score': 0.99}]

Popular Tasks & Models to Try

Text Generation

meta-llama/Llama-3.1-8B

Code Generation

deepseek-ai/deepseek-coder-6.7b

Translation

Helsinki-NLP/opus-mt-en-fr

Summarization

facebook/bart-large-cnn

Image Classification

google/vit-base-patch16-224

Speech to Text

openai/whisper-large-v3

💡 Pro Tips

Filter models by license: Apache 2.0 or MIT if you need commercial use rights
The Serverless Inference API is free for low-volume use — perfect for prototyping
Spaces are free to browse and test — no account needed for most public demos
GGUF format models on HF are optimized for Ollama — search for "GGUF" to find them
Use the HF Leaderboard (huggingface.co/open-llm-leaderboard) to compare open-source models before downloading

→ Run HF Models Locally with Ollama → Free Local AI Guide