@Humantoday_ i think it is reasonably accurate and yet it is going on just a generic prompt with the name "mimi10v3" - whatever it is drawing on is in its training data already