Latte macchiato
@latte
@plasmatrap.net
1y
@micr0
@bottie
interestingly 4o performs worse than the Google models at vision stuff
