This study aimed to analyze the effectiveness of the Somatic, Auditory, Visual, Intellectual (SAVI) learning and Problem-Based Learning (PBL) models in improving students' mathematical communication ability compared to direct learning. A quasi-experimental design was employed, with the SAVI learning model implemented in experimental class 1, PBL in experimental class II, and direct learning in the control class. The sample was selected using cluster random sampling, and data were collected through observations, interviews, and tests. Hypothesis testing, including proportion tests, one-way ANOVA, and Scheffe tests, revealed the following results: (1) students taught with the SAVI learning model demonstrated significant improvement in mathematical communication ability; (2) students instructed using the PBL model also exhibited substantial enhancement in mathematical communication ability; (3) notable differences were observed among the SAVI learning, PBL, and direct learning models, highlighting the effectiveness of both the SAVI learning and PBL models in enhancing mathematical communication ability. Moreover, the PBL model outperformed direct learning, while the SAVI learning model showed greater effectiveness than direct learning. Overall, this study provides evidence supporting the efficacy of the SAVI learning and PBL models in fostering students' mathematical communication skills.