the current will vary for sure.
But I would not assume, that it varies with the same “sleeping” software, but at slightly different voltage levels. At least not in that range.
I setup a second box to run your tools and projects closest possible.
I use https://github.com/circuitdojo/nrf9160-feather-examples-and-drivers.git in version 1.9.1 , I build the “sample/active_sleep” (the patch seems not longer matching nor required for v2.7.99-ncs1-1).
Using the Nordic PPK II in power source mode (GND and VBAT), I get different results (I guess, the sample is not completely up to date with ncs 1.9.1):
3.6 V => 200µA
4.0V => 200µA
4.1V => 470µA
Also a test with my own app (using ncs 2.1.2) results in a similar behavior (values of my first comment). It seems to have a “significant consumption step”, if 4.1V is reached and exceeded.