There was one reported observation in this forum that the 4900 showed more color inconstancy with changing light ("metamerism") compared to a 7890 printer output.
You mean metameric failure don't ya? :~)
I tend to discount that one observation as an outlier...I've printed with every Epson printer since the 10000 and can tell you that with each round of head and ink improvements, metameric failure is continually reduced. I don't see any metameric failure with my 9900 and I have several viewing booths to use for comparison. I would also note that gloss differential also seems better with the 9900 at least on the papers I print on; EFP, Luster, Semi. GD isn't an issue with fine art paper.
I would also say that Epson has tended to understate the usefulness of O & G for photos...originally designed for proofing to hit spot colors, Epson was actually worried that the added O & G would make it more difficult to profile (they had the same worry back when they released the R800 with red and blue inks). But I don't think the 9900 is any more difficult to produce profiles for than any other printer. The new dithering produced in association with Munsell and the internal separations from input to Cc, Mm, Y, K1 K2 K3, O, G seem very reliable for both photo and proofing as long as you print at 1440 or above (there is an issue when printing out at 720 but I won't get into that here).