Obesity has become a major global public health concern, affecting more than one billion individuals worldwide. As a low-grade chronic inflammatory condition, obesity is closely associated with cardiometabolic disorders and gut microbial dysbiosis. Diet-based interventions are recognized as one of the safest and most effective strategies for long-term weight management. Increasing evidence indicates that specific dietary patterns can modulate gut microbiota (GM) composition and metabolic function. However, comparative evidence regarding the effects of different dietary strategies remains limited and inconsistent. This systematic review and meta-analysis comprehensively evaluated the effects of the very-low-calorie ketogenic diet (VLCKD), Mediterranean diet (MD), and intermittent fasting (IF) on gut microbiota in obese populations. Systematic searches of PubMed, EBSCOhost, Cochrane, and Web of Science were conducted up to September 2025. Meta-analyses using R software assessed changes in microbial diversity and characteristic taxa abundance, with subgroup analyses by body mass index (BMI), age, and intervention duration. A total of 42 studies were included. Random-effects meta-analysis revealed that VLCKD significantly increased the Shannon index, observed OTUs, and Faith"s phylogenetic diversity (PD), promoted Akkermansia abundance and the Firmicutes/Bacteroidetes (F/B) ratio, but reduced Bifidobacterium abundance, indicating a bidirectional regulatory effect on gut microbial structure. MD significantly increased the Shannon index as well as the abundance of Akkermansia, Bifidobacterium, and Bacteroidetes, while decreasing Firmicutes abundance and the F/B ratio, suggesting a balanced and sustained improvement in gut microbial composition. In contrast, IF significantly decreased the PD index while increasing Akkermansia and reducing Firmicutes, reflecting partial structural optimization but limited enhancement of phylogenetic diversity; long-term interventions were associated with a decline in Shannon diversity, indicating limited stability. Subgroup analyses revealed distinct moderator effects. Under VLCKD, improvements in microbial diversity were more pronounced among individuals with BMI≤30 kg/m² and those aged 30 years, and meta-regression confirmed that the magnitude of diversity gains increased with age. Regarding BMI, increases in Akkermansia abundance were most evident in individuals with BMI 30–35 kg/m², whereas Bifidobacterium abundance significantly decreased in the same range, suggesting a threshold-dependent microbial response to adiposity. With respect to age, both Akkermansia (increase) and Bifidobacterium (decrease) exhibited significant changes in individuals aged 40 years. In terms of intervention duration, Akkermansia increased significantly within 6 weeks, while Bifidobacterium decreased within 12 weeks. For MD, increases in Shannon diversity were consistently observed across all BMI, age, and duration subgroups; notably, Akkermansia abundance increased significantly among participants with BMI30 kg/m², aged 30–50 years, and during interventions ≤6 months, while Bifidobacterium abundance rose markedly in participants with BMI≤30 kg/m², aged 40–50 years, and during interventions of 6–12 months. Under IF, Shannon diversity increased significantly in individuals with BMI≤30 kg/m² and aged 40 years but declined when the intervention exceeded 4 weeks, suggesting reduced long-term stability. In conclusion, VLCKD, MD, and IF all modulate gut microbiota in obesity but differ in magnitude, direction, and durability. VLCKD exerts strong yet dual effects—enhancing diversity while reducing beneficial taxa; MD shows stable, sustained modulation; whereas IF offers selective improvements but lowers phylogenetic diversity with limited persistence. Future studies should conduct large, multicenter randomized controlled trials to determine optimal intervention duration, confirm the moderating roles of age and BMI, and develop personalized microbiota-based dietary strategies for obesity management and gut health improvement.