# HuggingGPT - Multi-Modal AI Playground

A sleek, dark-themed Gradio app that brings together powerful language models, image generation, and text-to-video — all through the Hugging Face Inference API.

## Features

- **Chat** - Talk to Qwen 2.5, Kimi K2, or any custom HuggingFace model
- **Image Generation** - Create images with FLUX.1-schnell or FLUX.1-dev
- **Text-to-Video** - Generate videos from text descriptions
- **Personality Modes** - Choose from Professional, Creative, Witty, Coder, or Research Assistant
- **Dark Theme** - Clean, modern interface with customizable aesthetics

## How to Use

1. Set your `HF_TOKEN` as a Space secret (for gated/private models)
2. Enter your name and choose a personality on the welcome screen
3. Use the sidebar to switch between Chat, Image Gen, and Text-to-Video
4. Start creating!

## Models

| Type | Models |
|------|--------|
| Chat | Qwen/Qwen2.5-7B-Instruct, moonshotai/Kimi-K2-Instruct, Qwen/Qwen2.5-0.5B-Instruct |
| Image | black-forest-labs/FLUX.1-schnell, black-forest-labs/FLUX.1-dev |
| Video | tencent/HunyuanVideo, Wan-AI/Wan2.1-T2V-14B |

## Setup

```bash
pip install -r requirements.txt
python app.py
```

Or deploy directly to HuggingFace Spaces with a **Gradio** SDK.