The research aims to investigate and evaluate the efficacy of different learning models on mathematics learning outcomes, specifically focusing on the students' van Hiele geometric reasoning capacity. The study employed a quasi-experimental design. The practice is carried out in Public Junior High Schools in Sukoharjo Regency, part of the Central Java Province in Indonesia. The utilized learning models were transformative learning, actual mathematics education, and direct instruction.  The study encompassed a total of 281 individuals, who were selected from 9 classes across three distinct institutions.   Data analysis is conducted with a two-way analysis of variance (ANOVA) approach. The research findings indicate that (1) the learning model has a positive and significant impact on mathematics learning outcomes; (2) the van Hiele level of geometric thinking ability has a positive and significant influence on mathematics learning outcomes; (3) there is a positive and significant interaction between learning models and van Hiele level of geometric thinking ability on mathematics learning outcomes; and (4) among the transformative learning, Real Mathematics Education, and Direct Instruction models, the transformative learning model yields the best mathematics learning outcomes.