Both are correct in ways.
The eye changes its shape so that when you focus on a point, all the diverging light from that point is collected and redirected onto (roughly) the same detector on the back of your eye. (as in image B) However, in theory, you only need to collect a single photon to "see" a point (this corresponds more closely to A). Very sensitive light detectors can see a single particle of light, and IIRC, so can the eye if it is very dark.
In reality however, most objects emit light in all directions, so in realistic situations, B is the better picture. It's important to note that there are "an infinity" (in reality a large but finite number) of rays emitted in a cone from every point. There is nothing special about the ray distinguished in A. If you blocked the top ray in B, you'd still see the point, it would be the same situation as A, but the light would just be bent by the eye. Each additional ray simply adds to the amount of light collected by the eye, making the image more clear.