Only one way leading to the virtual Babylonian Tower of Pisa in Rome

[DNS-Leo]

I found a GPS receiver with an amazing precision up to 7mm.
I want two, collect their data, and calculate the free-sight straight-line distance between them.

Unfortunately I didn't receive too much math eduction, but Pi and Pythagoras I do understand.
Now English isn't my native language, plus all the overbearing geo terminology, is pretty difficult to me.
However, I do understand earth is not flat.

This is the code I have now:

from math import sqrt
from pyproj import Transformer

# 3km in Rome (according to Google Maps) + 4km altitude, to make a "3/4/5ish" triangle:

lon1=12.5
lon2=12.488
lat1=41.94215768476
lat2=41.91667795833
alt1=4000
alt2=0

gps2xyz = Transformer.from_crs('EPSG:4979', 'EPSG:4978', always_xy=True)

pt1 = gps2xyz.transform(lon1, lat1, alt1)
pt2 = gps2xyz.transform(lon2, lat2, alt2)

distance3d = sqrt((pt1[0] - pt2[0])**2 + (pt1[1] - pt2[1])**2 + (pt1[2] - pt2[2])**2)
print("distance 3D: %s m" % distance3d)

One approach to check the outcome, was using the central-angle-of-a-circle (earth) to calculate the never-exactly-90deg-angle + altitude + the length-of-the-non-curved-side, to find the hypotenusa (which would be the distance between my two gadgets).
That outcome came very close.

Still I wonder if this formula to calculate distance between 3D points is entirely correct applied for this data - or am I mixing two different worlds?

And if it's correct, then still, is this the most efficient and accurate way to accomplish the thing I'm trying?