Round 15 was in February of this year.
Referencing this previous topic: TechEmpower performance benchmark results round 15, including several play2 setups
Main point from the commentary: a lot of the environments(and thus their tests) have been ported to Docker containers.
New hardware, including a 10-gigabit Ethernet switch was used and it was gave some interesting observations. Feel free to read the blogpost.
As for Play Framework and its results:
Round 15 and Round 16 both used version 2.6.9 of the framework. Round 16 used scala 2.12.5 instead of 2.12.4 in round 15.
Apart from that all settings seem identical across versions.
The resulst seem weird. Despite framework versions not having changed, some categories improved(single query) others got worse(multiple query) json serialization.
Either way, those versions can stand to be updated. I’ll try and look into that.