min(DALL·E) is a stripped-down PyTorch implementation of the DALL·E Mini/Mega text-to-image model. It removes unnecessary dependencies beyond numpy, requests, pillow, and torch, and optimizes for inference speed. The repository provides a Python interface for loading model weights and generating image grids from text prompts, with performance benchmarks showing generation times ranging from 55 seconds on a T4 to 15 seconds on an A10G.