This study was motivated by the low speaking skills of eighth-grade students at SMP Teknologi Pekanbaru, as indicated by students’ lack of confidence in expressing opinions, low speaking fluency, and minimal participation in classroom discussions. Speaking skills require the ability to convey ideas orally in a coherent, clear, and confident manner; however, in reality, the learning process is still dominated by conventional teaching methods, resulting in unequal speaking opportunities among students. This study aimed to determine the implementation of the Cooperative Learning Time Token model, the differences in students’ speaking skills before and after the treatment, and the effect of the model on students’ speaking skills. The research employed a quantitative approach with a quasi-experimental design using a one-group pretest–posttest model. Data collection techniques included speaking performance tests and observation sheets of teacher and student activities. Data were analyzed using descriptive and inferential statistics, including normality tests, homogeneity tests, paired sample t-tests, and simple linear regression analysis. The results showed that the average pretest score was 67 and increased to 84.83 in the posttest, with an improvement of 17.83 points in the high category. The significance value of the t-test was 0.001 < 0.05, and the coefficient of determination was 0.873, which is categorized as very strong. Thus, it can be concluded that the Cooperative Learning Time Token model is effective and has a significant effect on improving the speaking skills of eighth-grade students at SMP Teknologi Pekanbaru.