Our group recently did a four-month deployment of SoundTraps in the northeastern Gulf of Mexico at depths ranging from ~100 to ~450 m. For the few instruments that were still recording when we recovered them (two ST500s and three ST600s), I was able to measure the exact clock offset and thus estimate their clock drifts (based on a linear drift assumption).
All the units were running slow, with drifts of 5.2 an 5.6 s/day for the ST500s and 7.7, 7.6 and 7.7 s/day for the ST600s. When we got back to the lab, I ran a deck test at room temperature on one of the ST500 and recorded a drift of ~1 s/day. Unfortunately, we don't have in-situ temperature measurements but it would be interesting to determine whether these really high drifts are consistent with temperature. Also, it's interesting that the ST600s seem to be experiencing a higher drift that the 500s, which are the older model.