Maybe the future is multiple models. You prompt an image and a router decides the best AI image model for that prompt… Same concept as Mistral 8x7B or Mix of Experts but with image models…
Router Architecture for Multiple Image Generation Models
By
–