Saying this as someone that used Afterstep and Windowmaker alongside GNUStep, and did seat a few times on the GNUStep room at FOSDEM.
Last time I checked was at the level of OS X Panther, and modern Objective-C still wasn't supported.
I’ve wanted to see GNUstep succeed, but unfortunately it never got as much attention as the KDE/Qt and GNOME/GTK ecosystems. I have some theories as to why, but I think the biggest barrier is those who really wanted OpenStep/Cocoa in the 1990s and 2000s could’ve used readily-available NeXT/Apple software instead of waiting for GNUstep. It’s the same issue ReactOS and Haiku have; they’re competing against Windows NT/2000/XP/Server 2003 and BeOS, respectively. Even FreeDOS, which is architecturally much simpler, took quite a while to reach version 1.0; people could just get MS-DOS 6.22.
Of course, the Linux kernel and the GNU ecosystem are counterexamples, though I believe it’s easier to reimplement Unix due to its modular nature than to reimplement entire GUI toolkits, especially if source- and/or binary-level compatibility are required.
A GNUstep that was ready around 1998 or 1999 to capture the attention of former NeXT developers and deliver ports of NeXT software to Linux would’ve been the ideal opportunity, though it still would’ve been quite an effort to bring over other things that made the NeXT special, such as Interface Builder. I’ve noticed that most commercial Mac OS X software in the early days were Carbon applications, not Cocoa applications. Many legendary NeXT software products did not make the transition from NeXTstep/OPENSTEP to Mac OS X. They could’ve had a home on Linux or one of the BSDs via GNUstep had GNUstep been ready.
I think it’s a safe bet there were many more Mac applications than NeXT applications (that’s one of the reasons Carbon was created), so I don’t see how that’s surprising.