I think that all of this is pure marketing and can be safely ignored. There is no basis for the generations of sequencing in chemistry or instrumentation, except possibly that second-generation (Illumina) was a massive parallelization of first-generation sequencing (Sanger) through clever chemistry and instrument modifications. It is merely "I want to argue that this technology occupies a particular niche right now".
All of these terms are obviously very imprecise. I had not heard "4th-generation" before; having analyzed all of these data types, I personally would lump nanopore into the third-generation sequencing box with PacBio: they are long, relatively noisy reads that grew popular around the same time for roughly the same applications and are currently under very active research and development.
I would allow that these two technologies (PacBio and nanopore) are definitely distinct from Sanger (much older, now fairly restricted to specific applications like confirming plasmid sequences) and Illumina (older, mature tech with deep market penetration, making tons of money and trying to hold onto market share).
"NGS" itself is a rather silly term for similar reasons, as it is now clearly a technology that is fighting for relevance in certain applications where it was once dominant (e.g. genome assembly). If we must apply generational terminology to Illumina, it is clearly second-generation. This is one of Joe Felsenstein's oft-repeated rants.
Nanopore itself is as far as I know established in 1996, whereas the first publication I am aware of that forms the basis for PacBio is 2003. For comparison, the Solexa sequencing that became the basis of Illumina instruments from thence was introduced in 2005. So any meaningful chronology also does not support these generations.