<p>Obsessive–compulsive disorder (OCD) is a highly heterogeneous neuropsychiatric disorder, marked by substantial individual differences in symptom expression, comorbidity, and treatment response. Converging evidence implicates dysfunction of the cortico-striato-thalamo-cortical (CSTC) circuit as a central feature of OCD pathophysiology. However, most existing research relies on symptom-based classifications or static neural markers, which are insufficient to capture the temporal evolution of symptoms and their state-dependent modulation under different interventions. Dynamical mechanisms underlying symptom persistence, fluctuation, and treatment-induced change remain poorly characterized. Dynamic modeling offers a mechanistic framework by formalizing behavior, cognition, and neural activity as interacting nonlinear systems, enabling systematic analysis of how pathological states emerge, stabilize, and transition over time. This review focuses on dynamic models of OCD, highlighting evidence for abnormal attractor stability, altered metastable dynamics, and impaired state transitions within the CSTC circuit, which may underlie compulsive rigidity and reduced behavioral flexibility. We further discuss how these models inform the understanding and optimization of pharmacological and neuromodulatory interventions, including selective serotonin reuptake inhibitors, deep brain stimulation, and transcranial direct current stimulation, by characterizing how such interventions reshape circuit stability and state-transition landscapes rather than simply suppressing neural activity. Overall, dynamic modeling provides a unifying perspective that links circuit dysfunction, symptom heterogeneity, and differential treatment effects in OCD, supporting the development of more precise, mechanism-informed psychiatric interventions.</p>