If one considers the possibilities of shifting quadra values relative to strong/weak functions, one has 64 types (or possibly 16*16 types if one tries to weight quadra values more individually all model A, which may be similar to Tcaud's dual-type concept).
I don't see how you arrive at that number.