Understanding coupled World-Earth system dynamics is central to navigating the Anthropocene, yet global social and environmental systems are still predominantly modelled separately. Bidirectionally coupled models of global scale that represent both subsystems as endogenous components are increasingly developed, but the field remains fragmented across disciplines, modelling traditions, and epistemological assumptions.</p> <p>This study presents a systematic review of global social-environmental modelling approaches, drawing on 314 peer-reviewed studies identified through a comprehensive database search and the usage of a newly developed, Large-Language Model assisted approach.</p> <p>We propose a typology of three model groups – Goal-Oriented models, Structural Feedback models, and Dynamic Co-evolutionary approaches – that can help to navigate the diverse landscape of this research field and assess them against the design principles for World-Earth Models proposed by Donges/Heitzig et al. (2020).</p> <p>Our analysis shows characteristic strengths and limitations tied to each typical model group. Across all, socio-cultural processes beyond economic optimization, diverse bidirectional feedback, endogenous tipping dynamics, and cross-scale interactions remain systematically underrepresented. A collaboration analysis further confirms the conceptional difference between the proposed groups and highlights the strong fragmentation of the field along the model groups.</p> <p>These findings suggest that future modelling efforts towards a better understanding of Anthropocene dynamics and World-Earth System resilience should prioritize broader actor representation, more diverse feedback structures, and explicit coupling of social and biophysical tipping processes, ideally through hybrid modelling formats – while remaining grounded in the specific research question rather than aiming for universal coverage.