Which is more efficient, raycast, get distance to x, or using collision?

So you feel that a big collision object (like say a cone from the character’s head) with the proper collision filters set would be more efficient than raycasting and checking distance? I do know that longer raycasts are less efficient.