Erdős conjectured that the true maximum is almost linear, roughly (n^{1+o(1)}), based on lattice examples giving many unit distances .
The best general upper bound known is still (O(n^{4/3})), proved by Spencer, Szemerédi, and Trotter in 1984 .
So the current situation is :
lower bound: n^(1 + O(1/log log n))
upper bound: O(n^(4/3))
conjecture: n^(1+o(1))
A simple example: put points in a square integer lattice; many pairs will be distance 1 apart horizontally and vertically, giving linear-in-(n) many unit distances, and Erdős’s lattice-based constructions give slightly more than linear many unit distances .