Score: consensus of the positions in 5 test sources, 100 = top of every source. Every model on this list is recommended by at least one test source. The bottom of a top list is still worth half the points, and a model named by only one source cannot score above 50. Only models sold in the United States count. Method.