The difference between the two, modelling wise, is that the sensors would need to be placed before the analysis and probes would be placed after the analysis.  The reason this is different is because for sensors, the information is tracked over every iteration of the analysis, whereas the probes can only capture data that was output through the intermediate time steps.

So, the advantage of sensors is that the data plotted over time will have a much greater resolution, but the downfall is that if the sensor needs to be moved, or if additional sensors need to be added, the analysis needs to be re-run.

For probes, the placement is much more flexible, and they can also be copied from run to run after a large group of analysis has been completed.  The downfall of probes is that the resolution of data plotted over time is completely dependent on how many time steps are output.