Abstract page for arXiv paper 2606.11289: i1: A Simple and Fully Open Recipe for Strong Text-to-Image Models