The most reliable way is to use a serial terminal program like MOSerial on Linux or HyperTerminal on Windows.
There you can query all parameters from the batteries and also the firmware versions.
There is supposed to be a mechanism where firmware updates are deployed from master battery to all slaves, but I found it more reliable to update each battery individually (I only have 2 of them) while they are disconnected from each other (no link cable, no nor power cables connected) and also disconnected from GX.
My Pytes 48100C are a little bit older than your’s, initially came with FW 1.5.18.C16 iirc, and I have seen a big improvement with 1.5.28 because they changed some BMS thresholds: ESS Merwürdiges Regelverhalten bei DC Überschusseinspeisung mit Pytes Akku - #17 by HansDampf
And one important thing to mind with these (and most likely other) batteries is that the upper 10% of the SoC (90..100%) as well as the lower 20% (0..20%) are hard to calculate / estimate and even harder, the less the cells are balanced, so it’s very normal to see jumps there because the SoC levels are calculated from voltage levels and as soon as a single cell is reaching a certain max or min threshold, the SoC may e.g. jump from 15 to 10% or 90 to 100%.
Attention: The battery FW version shown on GX is awkward - Pytes does not transmit the full version number (like v1.5.28.C16) but some stripped down version (v28.3 in my case).
My link above is also showing cell voltages towards 100% SoC. You won’t see 100% with only 3.45V. Balancing (passive!) starts at 3.36V and 100% is reported by the BMS as soon as a single cell of a module has reached 3.55V.