Penelitian ini bertujuan untuk mendeskripsikan: (1) perbedaan kemampuan memecahkan masalah antara siswa yang belajar menggunakan model problem based learning (pbl) dengan siswa yang belajar menggunakan model pembelajaran langsung; (2) perbedaan kemampuan memecahkan masalah antara siswa yang memiliki kemampuan akademik sedang dan siswa yang memiliki kemampuan akademik tinggi; (3) interaksi antara model problem based learning (pbl) dengan model pembelajaran langsung pada siswa yang memiliki kemampuan akademik sedang dan kemampuan akademik tinggi terhadap kemampuan memecahkan masalah. jenis penelitian yang digunakan yaitu eksperimen semu dengan desain penelitian post-test only control group design faktorial 2×2. populasi penelitian ini adalah seluruh siswa kelas v sdn 006 tarakan. pengambilan sampel menggunakan teknik cluster purposive sampling. teknik analisis data yang digunakan meliputi uji normalitas, uji homogenitas, dan uji hipotesis. uji hipotesis menggunakan uji mann-whitney test dan uji kruskal-wallis test. hasil penelitian menunjukkan bahwa (1) tidak terdapat perbedaan kemampuan memecahkan masalah antara siswa yang belajar menggunakan model problem based learning (pbl) dengan siswa yang belajar menggunakan model pembelajaran langsung dengan nilai asymp. sig. (2-tailed) sebesar 0,274 > 0,05. (2) terdapat perbedaan kemampuan memecahkan masalah antara siswa yang memiliki kemampuan akademik sedang dan siswa yang memiliki kemampuan akademik tinggi dengan nilai asymp. sig. (2-tailed) sebesar 0,005 < 0,05. (3) terdapat interaksi antara model model problem based learning (pbl) dengan model pembelajaran langsung pada siswa yang memiliki kemampuan akademik sedang dan kemampuan akademik tinggi terhadap kemampuan memecahkan masalah dengan nilai asymp. sig.(2-tailed) sebesar 0,026 < 0,05. kata kunci: model problem based learning (pbl), kemampuan memecahkan masalah

The objectives of this study were to describe: (1) differences in problem-solving ability between students who learn to use the problem based learning (pbl) model and students who learn to use the direct learning model; (2) the differences in problem-solving ability between students who have moderate academic abilities and students who have high academic abilities; (3) the interaction between the problem based learning (pbl) model and the direct learning model for students who have moderate academic abilities and high academic abilities on problem solving abilities. research employed a quasi-experimental research design with post-test only control group 2×2 factorial design. the population of this study were are fifth grade students at sdn 006 tarakan. sampling by using cluster purposive sampling technique. data analysis technique was use to describe normality test, homogeneity test, and hypothesis testing. hypothesis testing using the mann-whitney test and the kruskal-wallis test. the result showed that (1) there was no difference in problem-solving abilities between students who learn using problem based learning (pbl) model and students who learn using direct learning model with the asymp score. sig. (2-tailed) of 0.274 > 0.05. (2) there were differences in problem-solving abilities between students who have moderate academic abilities and students who have high academic abilities with asymp scores. sig. (2-tailed) 0f 0.005 < 0.05. (3) there was a good interaction between the problem based learning (pbl) model and the direct learning model for students who have moderate academic abilities and high academic abilities on problem solving abilities with asymp scores. sig (2-tailed) of 0.026 < 0.05. keyword: problem based learning (pbl), problem-solving ability