With little better to do in the current situation, I performed a wholly unscientific experiment a couple of days ago. On day one I rode a short local route on my eMTB, on the lowest assistance setting (which is the setting I normally use). On the second day I rode almost the same route (I forgot the final short loop around the local park, but did include the stiffest climbs) on my Tricross.
My Garmin reported that on the eMTB I covered 8.34km in 28m1s and used 281 calories in doing so. On the Tricross I covered 7.87km in 28m5s and used 319 calories. So, very roughly, on the eMTB I used ~17% fewer calories per km than on the unassisted bike, and my average speed was ~6% higher. My peak heart rate on the eMTB was ~4% lower, and my average heart rate ~7% lower.
I'm not sure whether I expected the assistance to have more of an effect than that. I may repeat the experiment a few more times to see how much variance I get in the results.
I suppose it might be interesting to try riding the route on the eMTB with the assistance turned off. "Try" probably being the approporiate word in that case - it's a heck of a lump compared to the Tricross...