Questions concering Gap in Output

Hello, I am currently comparing the results of my compact model with those of my column generation heuristics for a final thesis. To do this, I solve the compact model with a time limit of 7200 seconds (2 hours) and then read the gap from the output. Then I calculate the model with the same model instance with the Column Generation Heuristic and calculate the (LP) gap for the LP relaxation. Since my model is a MIP, this gap is actually always \(>0\), although the optimal IP solution may be found.
Now the question, the gap from the output is the (IP) gap or? How can I make the two comparable somehow? So the LP gap and the MIP gap? And is it possible to show that the column generation solution is also the optimal IP solution without having to solve the compact model completely?