Stable Diffusion Turbo XL can generate AI images as fast as you can type (arstechnica.com)

This magazine is from a federated server and may be incomplete. Browse more on the original instance.
IMO the only private LLM is one being run locally. Thoughts?