🤖 OpenAI GPT-OSS 20B - Run OPEN AI Locally!

Finally, truly OPEN AI! Run OpenAI's powerful GPT-OSS 20B model locally on your machine with a beautiful Streamlit UI. One-click installation, no cloud required, 100% private.

✨ Features

• 🚀 One-Command Installation - Get up and running in minutes
• 💻 Beautiful Web UI - Clean Streamlit interface for chatting with GPT-OSS
• 🔒 100% Local & Private - Your data never leaves your machine
• ⚡ Optimized for Apple Silicon - Leverages Metal Performance Shaders
• 📊 Real-time Stats - Monitor performance, response times, and memory usage
• 💾 Chat History - Export and save your conversations

📋 Requirements

• macOS (Apple Silicon M1/M2/M3/M4 or Intel)
• 16GB+ RAM (36GB recommended for best performance)
• 14GB free disk space for the model
• Python 3.9+ (usually pre-installed on Mac)

🚀 Quick Start

Step 1: Clone the Repository

git clone https://github.com/yourusername/ollama-gpt-oss.git
cd ollama-gpt-oss

Step 2: Run the Setup Script (One-Time)

chmod +x setup.sh
./setup.sh

This script will automatically:

• ✅ Install Homebrew (if needed)
• ✅ Install Ollama
• ✅ Download the GPT-OSS 20B model (13GB)
• ✅ Create Python virtual environment
• ✅ Install all dependencies
• ✅ Verify everything is working

Note: The model download may take 10-30 minutes depending on your internet speed.

Step 3: Start the UI

./run_ollama_ui.sh

Then open your browser to: http://localhost:8501

That's it! You're now running GPT-OSS 20B locally! 🎉

💬 Usage

Web UI (Recommended)

After running ./run_ollama_ui.sh, you can:

• Chat with the model in real-time
• Adjust temperature and max tokens
• Export chat history
• Monitor performance metrics

Command Line

ollama run gpt-oss:20b

Python API

import requests

response = requests.post('http://localhost:11434/api/generate', 
    json={
        "model": "gpt-oss:20b",
        "prompt": "Hello, how are you?"
    })
print(response.json()['response'])

📁 Project Structure

ollama-gpt-oss/
├── setup.sh                   # One-time setup script
├── run_ollama_ui.sh          # Start the Streamlit UI
├── ollama_streamlit_app.py   # Streamlit application
├── requirements.txt          # Python dependencies
├── ollama_api_example.py     # API usage examples
└── README.md                 # This file

🛠️ Troubleshooting

Model Download Issues

If the download is interrupted, simply run:

ollama pull gpt-oss:20b

Ollama will resume from where it left off.

Port Already in Use

If port 8501 is busy, you can specify a different port:

streamlit run ollama_streamlit_app.py --server.port 8502

Memory Issues

GPT-OSS 20B requires significant RAM. If you experience issues:

• Close other applications
• Consider using a smaller model: ollama pull llama3.1:8b

🔧 Manual Installation

If you prefer to install components manually:

1. Install Ollama:

brew install ollama
brew services start ollama

2. Download the model:

ollama pull gpt-oss:20b

3. Setup Python environment:

python3 -m venv ollama_ui_env
source ollama_ui_env/bin/activate
pip install -r requirements.txt

4. Run the UI:

streamlit run ollama_streamlit_app.py

🎯 Performance

On Apple M4 Max (36GB RAM):

• Model Load Time: ~5 seconds
• Average Response Time: 2-3 seconds
• Memory Usage: 8-10GB when active
• Token Generation: ~30-50 tokens/second

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• OpenAI for releasing GPT-OSS as open-source
• Ollama team for the excellent local LLM runtime
• Streamlit for the amazing web framework

⭐ Star History

If you find this useful, please consider giving it a star!

📮 Contact

For issues, questions, or suggestions, please open an issue on GitHub.

---

Made with ❤️ for the open-source community