Input patterns to the olfactory bulb are dynamic and change in an odor-specific manner, as measured by selective calcium imaging of olfactory bulb input. To our knowledge, none of the published models of olfactory bulb function uses dynamic input patterns. Therefore we tested how dynamic input alters the behavior of a simple model consisting of two layers. The membrane potential of the first layer neurons, integrate and fire neurons corresponding to mitral cells, was modulated with a sub-threshold oscillation at respiration frequency. The membrane potential of the second layer neurons was used to discriminate input patterns. We implemented oscillating input with amplitudes and latencies different for each mitral cell. Not only varying the input amplitudes, but also de-synchronizing the input, and varying the relation between latency and input amplitude, individually changed the model's performance significantly. The discrimination time was affected more easily than the number of second layer neurons that can differentiate an odor pair. Increasing the de-synchronization, i.e. the spread of latency values, reduced the differences in response time between strong and weak stimulus pairs without reducing the number of reacting cells. Input phase relative to the sub-threshold oscillation altered the effect of de-synchronization. Thus, dynamic input changes performance parameters of models of olfactory information processing that can be verified experimentally.