This study aims to cluster Indonesian regencies and cities based on chili price movement patterns using a model-based Multivariate Time Series Clustering (MTSClust) approach. Daily price data for four chili types: big red chili (cabai merah besar), curly red chili (cabai merah keriting), red bird’s eye chili (cabai rawit merah), and green bird’s eye chili (cabai rawit hijau), were collected from 72 regions between January 2022 and December 2024. Missing values were imputed using the Last Observation Carried Forward (LOCF) method. Price dynamics for each region were modeled using Vector Autoregressive with Differencing (VARD), producing coefficient matrices that captured temporal and cross-variable relationships. These matrices served as input for clustering, which was performed using six scenarios combining K-means and K-medoids algorithms with three distance measures: Euclidean, Squared Euclidean, and Canberra. Evaluation using Root Mean Square Standard Deviation (RMSSTD) and R-Squared (RS) identified the K-means algorithm with Canberra distance as the best-performing method, constantly has lower RMSSTD and higher R-Squared, with an average RMSSTD of 51.02559 and an average R-Squared of 0. 998623.