The 6AS7G is very different from the 5998 electrically. They are similar enough that they can often be substituted without modification to the gear, but if the gear is not designed for the 5998, it might not work properly.
A run down on the differences:
- The μ (mu: amplification factor) of the 5998 is 5.4, compared to the 2 of the 6AS7G, meaning that the same bias will result in different plate voltages. In a cap-coupled amp, this is not a big deal, but whenever there's DC coupling, this could be a problem, causing distortion or clipping, or even a release of the magic smoke if the amp is designed in such a way that it would be damaged by a change like that. I wouldn't worry too much about tube gear being damaged, but a hybrid might not like it. It depends on the amp.
- The Rp (plate resistance) of the 5998 is 350 ohms, vs the 6AS7G at 280. Not a huge difference.for the most part, it won't matter.
- The Gm (transconductance) of the 5998 is 15.5mS, vs 7mS. This is the most important difference for a cathode follower, which is the output stage topology of most headphone amps, like the Crack. The output impedance of a basic cathode follower is approximately the reciprocal of the transconductance. The output impedance of a 6AS7G-based cathode follower is 1/0.007=142.9 ohms, give or take. A 5998 cathode follower is approximately 1/0.0155=64.5 ohms. This is why 5998's in an OTL amp sound much better than 6AS7G's with lower impedance headphones; a better damping factor.
- The plate curves of the 5998 are mugc better (more linear) than the 6AS7G, so there is slightly less distortion.
An A2293 would be an interesting option to try, but it'd need an adapter. Same for the 7233. Both are obscure, cheap, single triodes in a 9-pin miniature package with very good performance. Based on specs alone, I would recommend a pair of 7233's over all of the other options, 5998 included.