SDXL API

Super Simple Implementation of the SDXL Pipelines from Diffusers as an API.

A high-performance, memory-efficient API for Stable Diffusion XL operations. Built for production use with zero file storage and automatic resource cleanup.

Features

• 🚀 Memory-efficient operation with zero file storage
• 🎨 Support for text-to-image, image-to-image, and inpainting
• 🔧 Multiple scheduler options
• 📦 LoRA model integration
• 🔄 Automatic device selection (CUDA/MPS/CPU)
• 🧹 Aggressive memory cleanup
• 🛡️ Production-ready error handling

For MacOS Silicon (M1/M2) users:

# As of the time of posting this README, Mac Silicone users should use this Nightly version of Torch for FP16 support on MPS.
pip install --pre torch --index-url https://download.pytorch.org/whl/nightly/cpu

The server will start on http://localhost:8000 by default.

Memory Management

The API is designed for long-running deployments with:

• Zero file storage
• Automatic cache clearing
• Aggressive memory cleanup
• Resource monitoring

License

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

Acknowledgments

• Stable Diffusion XL
• FastAPI
• Diffusers