Well...
* TI gave up on the TI-68k series nearly 17 (89, 92+) and 15 (V200, 89T) years ago;
* brand-new TI-68k calculators have always been, and are still, sold at high price tags, and as such, price tags for used calculators aren't necessarily as low as they could be;
* no TI-68k model has a PTT mode or a PTT LED;
* the 68000 is relatively unpopular nowadays. It shows in the toolchain: it's been over a decade since m68k stopped being a tier 1 GCC ISA, with optimization bugs such as
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=40454 creeping in, and correctness issues becoming more frequent (
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88589 , though I didn't use upstream GCC). GCC4TI's heavily patched GCC 4.1.2 doesn't build, let alone produce correct code, on modern computer OS and toolchains;
* neither NewProg, nor GCC4TI did budge the apathy of the community, which largely died in 2006-2007, effectively when the Nspire was introduced but clearly not because of it.
I could certainly go on, but I think you (plural) are getting the point very well. Sadly, as much as I like, and I am / was knowledgeable about, the TI-68k series, it's been over 11 years since I understood that nothing would make the TI-68k series popular again. I realized it several days after the initial GCC4TI release. It's probably the first time I'm stating it in a public manner on one of the message boards of the community, although I already stated it on unlogged IRC chans.
Amusingly, tonight, I spent like 10 minutes, for the first time in 2020 and AFAICT for the first time in nearly a year, working on GCC4TI, specifically doing a copy-paste-modify job adding support for MIN_AMS=102 to GCC4TI, now that we know that this version is "publicly" available (well, in a prototype calculator which mr womp womp bought). I'm not even convinced I'll publish the commit: nobody will use it and it's trivial anyway. But working on GCC4TI is an exception, not a rule. Ever since I became the maintainer of libti*/gfm/tilp & skinedit/tiemu, shortly after being one of the creators (and the main developer) of GCC4TI, I worked more on the LPG software, even though I was less familiar with them, simply because they have more users and are therefore more useful.
Frankly, at this stage, IMO, and even if it has its flaws, the NumWorks N0110 is a much better development platform than the TI-68k series
People don't even need a real NumWorks calculator to develop for this platform: I've contributed to the Epsilon firmware, in both N0100 and N0110 times, without owning either calculator. I only saw a real NumWorks calculator (and a real Casio Graph 90+E, too) like twice, when meeting JB Boric + Adriweb + Bernard Parisse, and Bernard Parisse alone.
EDIT: the openness of the TI-eZ80 platform was TI's best protection against some developers fleeing to the open and technically globally (much) superior NumWorks platform. That won't happen any longer... though who knows, some stupid standardized testing regulators (a.k.a. enemies of mankind progress) might find a way to force closing down the NumWorks platform just like the TI-eZ80 series, and thereby kill the NumWorks company, reducing the number of modern graphing calculator manufacturers back to the three big entrenched manufacturers...