Eh...I may be forgetting a Sony PR, but I believe it's their fans who have been claiming 15 stops DR.
Every since the D800 came out fanboys have been acting as if ALL Sony sensors could deliver 15 stops (rounding up) of DR. This was never the case. The D8x0 series always stood out as the highest DR bodies, with a few other Nikon sensors delivering similar performance, all of them in the 14-15ev range. Most Sony sensors have been at or below the 5D4 (i.e. 13-14ev range). There are some notable exceptions where a Nikon or Sony sensor delivers 12s.
These same people tend to assume that ALL Canon sensors have the DR of a 5D3 with banding or a 6D2 with harsh shadow noise.
As I recall the mkII A7's did lose DR compared to the mkI's, and then the mkIII's actually got close to the D8x0 series. I don't feel like looking at DxO at the moment** to get exact numbers, but I think the mkIII's are on par with the D8x0 series, all of which are close to 15 (if you round up to a whole integer).
I've admittedly been a broken record regarding DR on this forum (and others) because...
- Most people really have no idea how a DxO DR score translates into reality, i.e. what it actually means for image processing. DR is just pushed on sites like dpreview so they get GAS and "know" they need it.
- Most images are not exposed/processed for maximum DR. Example: if you have a D850 and you're not using ETTR then you're not getting any more DR than a 5Dsr shooter who is using ETTR. If you're shooting JPEG you're not doing any better than cameras more than a decade old in RAW.
- 1ev in the DxO score amounts to little more than a NR slider adjustment. 2ev means a real shadow detail difference that cannot be eliminated with post processing technique (NR), but only in images pushed to the max in terms of DR. 3ev and more starts to equate to actual difficulties with wide DR scenes that can only be shot in 1 frame.
** I really don't ever feel like looking at DxO, but their DR tests are internally consistent and reasonably close to reality. The rest of their tests are demonstrably trash.