Consumer Price Index (CPI) is an index to measure the average change in the prices level of a basket of consumer goods and services consumed by the households in a certain period. CPIs in nearby cities may correlate each other. Therefore, there are location effects along with time effect for CPIs series. Generalized Space Time Autoregressive (GSTAR) model can be employed for forecasting CPIs that involves time and location information. In addition, there are exogeneous variables that affect CPIs series. Such variables are called input series. In this research, money supply was considered as input series. Therefore, the GSTAR model was extended into GSTAR with Exogenous Variable (GSTARX). We applied the GSTARX model for forecasting CPIs in four cities in Kalimantan: Pontianak, Banjarmasin, Samarinda and Balikpapan. The empirical results exhibited that GSTARX with normalized cross-correlation weight resulted in white noise residual with minimum Akaike?s Information Criteria (AIC) value.