We present a computationally efficient strategy that allows us to simulate magnetization switching driven by spin-transfer torque in magnetic tunnel junctions within a micromagnetic model coupled with a matrix-based nonequilibrium Green's function algorithm. Exemplary simulations for a realistic set of parameters are carried out and show switching times below $4\phantom{\rule{4.pt}{0ex}}\text{ns}$ for voltages above $300\phantom{\rule{4.pt}{0ex}}\text{mV}$ or around $2\ifmmode\times\else\texttimes\fi{}{10}^{10}\text{A}\phantom{\rule{0.16em}{0ex}}{\text{m}}^{\ensuremath{-}2}$ for the $\text{P}\ensuremath{\rightarrow}\text{AP}$ (parallel-to-antiparallel) direction. For $\text{AP}\ensuremath{\rightarrow}\text{P}$ switching, a trend reversal in the switching time is seen, i.e., the time for magnetization reversal first decreases with increasing bias voltage but then starts to rise again.