This study discusses the development of a deep learning-based instructional design model as an effort to improve students' critical analytical skills in Islamic Religious Education (PAI). The study is motivated by the fact that PAI learning practices are still largely focused on rote memorization (surface learning), resulting in students' critical, reflective, and contextual thinking abilities not being optimally developed. Therefore, this research aims to develop and analyze the effectiveness of a deep learning-based instructional model in enhancing students' critical analytical skills, particularly in the aspects of interpretation, evaluation, and argumentation. This research employed a Systematic Literature Review (SLR) approach by examining national and international scholarly articles indexed in Google Scholar, SINTA, and GARUDA published between 2020 and 2025. The findings reveal that the implementation of deep learning is able to improve conceptual understanding, students' active engagement in learning, learning motivation, creativity, and critical thinking skills. In addition, this approach helps students internalize religious values more deeply and meaningfully. The deep learning approach is considered relevant to the demands of 21st-century learning and the implementation of the Merdeka Curriculum because it emphasizes active, reflective, and contextual learning processes. This study contributes to the development of a PAI instructional design that integrates cognitive, affective, and spiritual dimensions, making it a potential innovative alternative for improving the quality of Islamic education in the modern era