Abstract: In order to promote the development of new eco-friendly insulation gases, the development of empirical and semi-empirical models for dielectric strength prediction of single and mixed gases were analyzed. The selection of molecular parameters is the basis of the empirical models. The introduction of the general interaction property function (GIPF) and the group contribution method can improve the accuracy of model. It is necessary to judge the applicability of the above-mentioned molecular parameters from the perspective of gas discharge so as to try to find or define particular parameters in the field of gas insulation. In addition, the gas discharge data should be standardized. The insulation strength of various gases should be evaluated under uniform test conditions, thereby the effect of sample quality on model error can be reduced. Due to the insufficient theoretical investigations on discharge, the empirical models should be used in conjunction with semi-empirical models. As for the mixed gas prediction model, the main difficulty lies in the prediction of synergistic effects, which may be related to the process of molecular interactions. The prediction model guided by the complete gas discharge theory may have higher accuracy and generality, which can further improve the efficiency of developing eco-friendly insulation gases.