Is Diffusion a supervised or unsupervised model?

Is Diffusion a supervised or unsupervised model? Is it different when it is a naive ddpm and when it’s a text to image model such as stable diffusion?

Neither. It’s a different process entirely.