Background: Although some researchers have applied both cooperation and competition in the practice of robotics education, most of these practices were not systematic, and there was no integration design for intra-group cooperation and inter-group competition. Objectives: The study aims to examine the effects of different coopetition designs on students' intragroup cooperation and intergroup competition. Methods: A two-stage study was conducted in two elementary schools. In the pilot study, two coopetition designs were developed by combining rewarded and unrewarded competition. In the follow-up study, three coopetition designs (i.e., zero-sum with real-time reward, zero-sum with delayed reward, and social-comparison) were examined. Results and Conclusions: The study showed that both social-comparison coopetition and zero-sum real-time reward coopetition can be more effective than zero-sum delayed reward coopetition in teaching knowledge and skills in robotics education. Social comparison coopetition turned out to have a larger effect on affective outcome than the zero-sum delayed reward coopetition. Implications: Social comparison coopetition seems to be a more advantages method for both cognitive and affective learning. However, more research is needed to confirm the difference in effectiveness between the three models of coopetition.