FLUX.1 and WAN were already miles beyond Stable Diffusion a year ago. While America has been pushing capitalistic goals with LLMs, China has been releasing banger model after banger model, all as open-source.

Downside is it's China and comes with their own biases, both with image generation and politics within its text models. Fortunately, some of that can be re-trained back to normal, but the technology is moving so fast.