Comment by Wowfunhappy

Claude is not very good at using screenshots. The model may technically be multi-modal, but its strength is clearly in reading text. I'm not surprised it failed here.