This study investigates the effectiveness of a Flipped Classroom combined with the Group Investigation (GI) method in improving mathematical communication skills among primary school students, viewed from the perspective of self-regulated learning and aligned with SDG 4 (Quality Education). A quasi-experimental 2×2 factorial design was applied involving 60 fifth-grade students from SDN 2 Citundun (experimental group) and SDN Citikur (control group) in Ciwaru District, Kuningan Regency, Indonesia. Instruments consisted of a mathematical communication test based on NCTM indicators (clarity of ideas, representation, reasoning, and oral communication) and an Self-Regulated Learning questionnaire. Data were analyzed using two-way ANOVA followed by Tukey’s post hoc test. Results showed a significant effect of the learning model on mathematical communication (F = 23.88, p < 0.001, η²p = 0.287), with the Flipped–GI group achieving higher posttest scores than the conventional class. Students with high Self-Regulated Learning outperformed those with low Self-Regulated Learning (F = 12.77, p = 0.001), confirming the role of learning independence. An interaction effect was also found (F = 4.35, p = 0.042), indicating that the Flipped–GI model benefited high-Self-Regulated Learning students the most; however, students with low Self-Regulated Learning in the experimental group still experienced meaningful improvement. The integration of digital pre-class learning and collaborative investigation in class provided more opportunities for explanation, representation, and argumentation. These findings demonstrate that the Flipped–GI approach effectively strengthens mathematical communication and supports inclusive, higher-quality learning, contributing to the achievement of SDG 4 in primary education. The model is recommended as an innovative instructional strategy for mathematics learning at the elementary level.