Abstarct: In recent years, the problem of small-scale regionalization has to be considered because of the need reliable statistics have been very much considered. The major problem here is the impossibility of measuring the target variable for each individual in the area in question. Even sampling from all areas under study could lead to high financial and time costs. Several statistical models have been proposed to achieve this goal. The main purpose of these models is the use of auxiliary information to upgrade direct estimates. To this end, the importance of using spatial information of areas within the frxamework of spatial models plays an effective role in analyzing small areas. The similarity of space between adjacent areas is also useful information that small space-baxsed models have been proposed to use for this information. In view of the fact that in many analyzes of small regions with non-normal responses such as numeric responses, we classify the generalized linear space mixing models in the frxamework of Bayesian inference for the analysis of small area data. By studying the simulation and analysis of the insurance data package in Gilan province, we show the performance of these models.