CoF-T2I: Video Models as Pure Visual Reasoners for Text-to-Image Generation Paper β’ 2601.10061 β’ Published about 1 month ago β’ 30