The place I see it most is in networking RFCs.
Beat me to it. "Octet" is used in a lot of networking documentation, and it also nicely decouples wire formats from any meaning the host system might have for the information. From IP's point of view, an octet is literally that: a bag of 8 bits, without meaning. It's not part of a floating point number, or a UTF character, or anything else but just plain undifferentiated data.
I see it used (and use it myself) from time to time in contexts where you want to differentiate between 7/8/9-bit bytes, mostly when working with embedded/low-level software.