Statistical Significance of a learning Model

#1
I built a learning model (for classification) based on a Random Forest classifier and i am asked to assess the statistical significance of its performances.

Up to now, i trained and tested it on two different datasets A and B, respectively.

What kind of test can i use?