In order to provide more meaningful runtime results, number of test cases for each question should be increased and each test case should be measured multiple times.

Sometimes a solution that takes 6ms is only faster than 50% of submissions and a solution that takes 5ms is faster than 100% of submissions. If these solutions are benchmarked many times each (e.g. 1000) and their average performance is measured instead, the difference in performance might become much less than 20% (5ms vs 6ms).

And in general, I suggest using JMH for measuring the runtime performance of submissions.