Mathematical problem-solving ability refers to the capacity to solve problems through sound decision-making, which is essential for effective learning. Students must be able to think positively, logically, and methodically to address mathematical problems. Based on local cultural knowledge, this study aims to examine how high school students taught using the PAKEM model (Active, Creative, Effective, and Fun Learning) differ in their mathematical problem-solving abilities compared to those taught through direct instruction. The population of this study comprised all Grade X students at SMA Negeri 1 Tanggetada, Tanggetada District, Kolaka Regency. The sample was selected using purposive sampling, consisting of 30 students each from classes X1 (experimental group) and X3 (control group) of the same school. This quasi-experimental study employed quantitative research methods. Data were collected using test instruments, and the analysis techniques applied included Ngain analysis, independent t-test, homogeneity test, and normality test. The research results show that the H₀ was rejected and H₁ was accepted, with a tcount of 5.239 and a significance value (2-tailed) of 0.000, which is less than the alpha level of 0.05. It indicates that the experimental class taught with the PAKEM model performed better in mathematical problem-solving than the control class taught through direct instruction. Qualitative analysis showed that three students' strong potential in solving mathematical problems through the application of the PAKEM learning model. The pretest and posttest scores show significant improvement: the experimental group increased from 57.90 to 85.13, while the control group improved from 56.57 to 78.13. These results indicate that the PAKEM model, grounded in regional local wisdom, has greater instructional effectiveness than direct instruction in improving students’ mathematical problem-solving abilities.