Based on the observations that have been made, it is shown that students' mathematical communication skills are still relatively low so that an innovative learning model is needed to improve students' mathematical communication skills. This study aims to see whether the improvement in students' mathematical communication skills taught with the Two Stay Two Stray cooperative learning model is higher than the mathematical communication skills of students taught with the Think Pair Share cooperative learning model and to see how the improvement in mathematical communication skills taught with the two models. This study is a study that is included in the quasi-experimental category with a pretest-posttest control group design. The subjects in this study were 62 students in grades VIII-A and VIII-E of SMP Negeri 4 Balige. Based on the research that has been done, the results of the N-Gain test were obtained with an average N-Gain value in experimental class 1, namely 0.813 in accordance with the Gain index criteria in the high category and the average N-Gain value in experimental class 2, namely 0.693 in accordance with the Gain index criteria in the medium category. Based on the research that has been done, because the data is normally distributed and homogeneous, the hypothesis test is carried out using the t-test. Through the t-test that has been done, it was obtained that , namely 2.465>2.000. Therefore  is rejected, which means that the increase in mathematical communication skills of students taught with the Two Stay Two Stray cooperative learning model is higher than those taught with the Think Pair Share cooperative learning model.