to be fair they really are not there yet. They are just in the "data collection" phase, the actual training and then tuning is still to do.

but hey, those are the same people who made the dataset (laion5b) for stable diffusion. I have hope.