I've had a similar issue to @William-Alpine. I usually look at ECMWF meteograms for specific locations, which display the elevation at the right. Here I've clicked on two nearby locations that differ by ~1500m of elevation. The meteogram shows almost identical temperature forecasts, because as the "model elevation" indicates, the model is (I imagine) being queried for the elevation that its much lower-resolution topographical model thinks the surface is at:

While I appreciate that, given enough knowledge, the user can figure out that despite the map taking up most of the page saying "2500m", and "Elevation" at the right saying "2500m", that the forecast is for the "model elevation" of 1500m, this seems to me a pretty high expectation for the user's knowledge. Further complicating matters, there is a slider for elevation/pressure that adjusts variables on the map, but doesn't change anything on the meteogram.

I don't know how feasible it is, but my understanding was that it's generally possible to query each model for the elevation that is shown on one's own (high-resolution) topographic model, such that every model is displaying the forecast for the same elevation. In my imagination this was how windguru, for instance, gives a comparison of different models at the same forecasted elevation, rather than different elevations for each different-resolution model.