This study aims to determine: (1) the learning outcomes of science taught by the time token type of cooperative learning model are higher than the learning outcomes of science taught by the think pair share type of cooperative learning model; (2) the learning outcomes of science students who have a high level of linguistic intelligence are better than those of students who have a low level of linguistic intelligence; and (3) the interaction between the use of cooperative learning models and linguistic intelligence on science learning outcomes. The population in this study were students in grades VA and VB at Pelangi Medan private elementary school, totaling 2 classes. The sample in this study was determined randomly using a non-probability sampling technique with the type of convenience sampling, with 26 students in each class. The research method uses quasi-experimental research (quasi-experimental). The instruments used in this research are the science learning outcomes test and linguistic intelligence scale of linguistic intelligence. Analysis of the data used is 2 × 2 factorial ANOVA. Based on the research results obtained: (1) The learning outcomes of students in the cooperative learning model class with the time token type are higher than the students' learning outcomes in the cooperative learning model class with the think pair share type with Fcount > Ftable (10,765). > 4.04); (2) Science learning outcomes of students who have high linguistic intelligence are higher in science learning outcomes of students who have low linguistic intelligence with Fcount > Ftable (14.994 > 4.04); and (3) there is no interaction between learning models and linguistic intelligence on science learning outcomes with Fcount < Ftable (0.079 < 4.04). The results are expected to provide information for teachers and schools to use the time token type of cooperative learning model to improve students' short story appreciation skills. In addition, it is necessary to improve students' linguistic intelligence in classroom learning.