Comment by freediver Comment by freediver 6 days ago 1 reply Copy Link View on Hacker News It uses a combination of LLMs, selecting the best output. (from the blog post)
Ah, I missed that. Thank you!