I did some tests with the tube (for now, without GMC yet).
Its sensitivity is good - it detects a small uranium drinking glass from 10cm.
Test #1: Background fluctuation (~15 min)
Test #2: I placed 2 radioactive objects of different intensity at ~10cm from the tube (CPM values have some delay because of the sliding 60s window calculation). Even from such a distance, it's detectable.
It works well, but I'm still not sure if it makes sense to connect a SI-22G tube to GMC, or just keep the tube for separate experiments.
The voltage is 390V, looks standard. A conversion rate is 792 to 1uSv/h (for M4011 it is 151) so it's about 5x more sensitive than a "standard" M4011.
I decided to use it with a Raspberry Pi, and upload the data to gmcmap with a software correction - no need to confuse people with a "red" point with high cpm values on a map :)
It's a weird decision that gmcmap uses CPM values as a default marker; uSv/h would be more universal.