The helicopter in question is a Syma S107G. It has two counter-rotating main rotors, and as a result, does not have an anti-yaw rotor. (It uses a small tail rotor in the same plane as the main rotor for pitch control.)

In a gradual ascent towards the ceiling, it will fly normally until it is within a very small distance - two or three inches - from the ceiling. At that point it will experience a sudden and dramatic uncommanded increase in vertical speed and smack into the ceiling.

Similarly, in a gradual descent, as few inches above the floor the decent will stop without a change in control inputs. This is the well-known ground effect - it takes less power to hover near the ground. I have always understood this to be a result of the rotor downwash creating a region of higher pressure air, which improves airfoil performance, all other things being equal.

But what's going on with the ceiling? Shouldn't it be the other way round - shouldn't the helicopter struggle to climb the last few inches?