It could well be the data used by the DWD is more recent or less recent then ours ? We get the GAFOR data pushed by DWD, and I suppose they do the same thing for their website. In EV4 you can do MapLayers->Weather->DWD GAFOR Full briefing. At the top of that page you see the validity date&times as given by the DWD. 

I guess its a matter of definition : if the GAFOR is valid from 1300-1500z, is 1500z then considered within the timeframe or not? Currently EV4 considers it outside the validity timeframe and does not display it at 1500z. 

ok, so the problem only exists for the last picture of any set.
The first and second pictures are changing to the next one on every odd hour.   But the last picture  should be displayed until the end of the validity period.

I have not checked, but the same applies to any wx forecast.  If Thunderstorms are forecasted until 1800, followed by CAVOK, then EV4 SHOULD display a Thunderstorm symbol for the 1800 hour.

Its not that simple. A WX data supplier gives forecast for at-the-hour timestamps. So suppose a WX forecast is given for T0, T1,T2 and T3, then 4 "forecast slices" exist.  so its not "until" a specific time is reached, its the forecast for the given timestamps. It does not tell if the situataion forecasted for 1600z is established at 1510z or 1550z. 

I kept an eye on the GAFOR forecast today in EV4, I noticed the from-to timespans for each forecast changes from a span of 3 to a span of 5 hours sometimes. EV4 showed forecast images for all timestamps, including the last one. This GAFOR data updates also infrequently. Sometimes we get 5 updates in half an hour for the same timespan (I can't check if the updates are just textual or also the graphical part), sometimes only one update per hour is received. I guess it all depends on the "dynamicness" of the weather at the time of creating the forecast.