Each category is one buying question, put to twelve models in six framings: 72 answers a category per segment. Share counts the four framings that ask for a pick, so 48 of the 72. The verdict comes first: whether the models agree, whether one product leads clearly, or whether the order is still contested. The page for each category shows who leads, who is warned against, how every model answered, and every judgment call made between the raw labels and the numbers.