The mechanical behavior of nanostructured metals is dictated by their microstructure, as well as characteristic timescales and lengthscales. Prominent examples are single crystalline pillars and nanoporous foams. Computational homogenization tools are useful for linking macroscopic response with microstructural behavior by means of representative volume elements of the system. FFT spectral methods offer improved numerical performance over standard Finite Element based homogenization, thus allowing for detailed RVE models. FFT homogenization for metals rely on the crystal plasticity model, which assumes the homogenization of dislocation ensembles at the nano-scale to define a flow-rule for each slip system. However, at sub-micron scales, the plastic behavior of metals is dominated by few discrete slip events of stochastic nature producing discontinuities along the slip planes involved. Therefore, standard crystal plasticity is not adequate for these sizes. In this work, an stochastic approach considering the slip of a collection of individual planes is implemented on an FFT based homogenization software by introducing each slip event as a shear eigenstrain field. The implementation is used first to predict the behavior of tungsten micropillars. The model is finally extended to the simulation of nanoporous W with random bicontinuous open-cell RVEs obtained using levelled-wave methods. The model results are compared against experimental observations. It is shown that both macroscopic stress-strain curves and deformation patterns agree with experimental results.