Hello,
my 3458A, which has a modified oven to 60°C instead of the nominal 95°C, starts at +9ppm after turn on, and settles to < 1ppm deviation after 10 minutes.
Maybe it then undershoots a bit , but after 2h it should be stable to < 0.2ppm, w/o ACAL.
Hard to tell, if your units have a problem, as an ACAL after warmup might give another big deviation, depending on the ACAL temperature done before storage and now, in your location.
Account for another 0.3ppm/K difference.
The Service Note 18A describes "Long term Drift", max. 15ppm, which is a hysteresis effect of the LTZ1000A itself, after a longer period of storage.
This is caused by the high oven temperature used, as far as we know.. at lower oven temperatures this effect usually does not occur. I further assume, that might be a basic problem for all 3458As, not for a certain batch of their references only.
Anyhow, it can be cured inside your instrument by six weeks of letting it run 24/7.
After it has stabilized to account for this drift, you may check the "Short Term Drift" of U180.
If you test the CAL72? gain parameter, don't forget to measure the internal temperature of the 3458A for each test point, as this gain is also depending on the ~ 0.25ppm/K T.C. of the U180 ADC chip.
This is not mentioned in the SN 18A, but might give a distorted picture, when your environmental temperature changes more than 1..2 °C.
Frank