logoalt Hacker News

Ey7NFZ3P0nzAetoday at 9:12 PM0 repliesview on HN

Well, behind "models" not "langual models".

Of course models purely made for image stuff will completely wipe it out. The vision language models are useful for their generalist capabilities