STEM project-based learning (STEM-PjBL) is widely promoted for developing creative thinking, yet evidence from Indonesian junior secondary classrooms rarely goes beyond class-mean gains, and renewable energy studies have concentrated on senior secondary students. This study examined changes in junior secondary students' creative thinking during a STEM-PjBL unit on renewable energy and described their responses to the approach. We used a pre-experimental one-group pretest–posttest design with one intact Grade VIII class (n = 28) in OKU Timur, South Sumatra. Over three sessions, groups designed, built, tested, and presented water-turbine, wind-turbine, and solar-panel prototypes. We measured creative thinking with an expert-validated open-ended test scored on fluency, flexibility, originality, elaboration, and STEM integration; an observation rubric and a student response questionnaire provided supporting evidence. We analyzed data using the Wilcoxon signed-rank test, effect sizes, and normalized gain. Mean scores rose from 47.26 (SD = 22.00) to 91.19 (SD = 4.95); 27 of 28 students improved (Z = −4.60, p < .001, r = .87; dz = 2.11). The class-average normalized gain was 0.83, with 24 students in the high-gain category. All five indicators improved, while STEM integration remained the lowest at posttest. Observed creative behavior and the class response profile were favorable. The findings suggest that prototype-centered STEM-PjBL can create favorable conditions for creative thinking in renewable energy learning, although the single-group design, posttest ceiling, and brief intervention call for cautious interpretation. The results support explicit scaffolding of mathematical and engineering reasoning within junior secondary project work and warrant controlled replication.