Mathematical modeling for permeability prediction of mixed matrix membranes (MMMs) is rapidly improving. Ideal models failed in most of predictions. Development of a precise model for permeability prediction of MMMs with considering interfacial defects and pore blockage of particles is accessible. In this study, a new pseudo two phase model considering effects of all interfacial defects and pore blockage of particles was developed. In the developed model, permeabilities of pseudo dispersed phase and whole MMM were calculated by Gonzo-Parentis-Gottifredi and Lewis-Nielsen models, respectively. Effects of interactions of particles, size distribution, shape and agglomeration of particles were considered in new developed model. Also a fast and comprehensive optimization algorithm for determination of model’s adjustable parameters was developed. This algorithm has the capability of considering simultaneous effects of polymer chain rigidification or nonselective voids around the particles interphase and pore blockage of porous particles. It benefits lack of usage of arbitrary values for adjustable parameters, optimization of interphase thickness within a logical range, and reduction on dependency of using experimental permeability data. The developed model and algorithm were validated with 12 sets of different experimental data (129 data points). Predictions of new developed model for different gases from these sets of data showed that, 55.6%, 33.4% and 11% of percentage of average absolute relative error values that obtained from these permeability predictions are included values of less than 5%, 5-10% and more than 10%,respectively. Also the results of proposed modified Maxwell model in this study showed that, 33.4%, 44.4% and 22.2% of values are included values of less than 5%, 5-10% and more than 10%,respectively. These results show the superiority of new model to proposed modified Maxwell model. Permeability predictions of the developed model were compared with 8 data sets of published modeling results. The comparison reveals the capability of the developed model in providing of more accurate prediction data regardless of using less experimental for the optimization process. The developed model and optimization algorithm for pure gases were modified for permeability prediction of each component of gas mixture passing through MMMs. This modified model predict the permeability of each gas component of a gas mixture without any need to experimental data of the gas mixture passing through filler particles. The modified model consider effects of interphase and pore blockage of filler particles, effects of interactions and competition between components of gas mixture on the interphase and pore blocked filler particles. Keywords : Gas separation, Mixed matrix membrane, permeability, modeling, interphase, particle pore blockage, optimization, pure gas, gas mixture