Why does Radius Solo Vocal introduce more artifacts than Radius Solo?

Use the one that sounds the best.

It might work better with individually moved transients as opposed to a full-clip linear stretch or it might just be something about the nature of the particular vocal that you're stretching.  I've mentioned before that I've often found the default Groove Clip algorithm to sound better than dedicated Radius algorithms for full-clip stretching.

After my last post, I tried four algorithms (Radius Solo, Radius Solo Vocal, Radius Mix and Groove Clip) on a slip-stretched full-length solo vocal clip in two different projects (I completely forgot to test the algorithms with shorter clips ), and, again, found that Radius Solo Vocal introduces more artifacts than Radius Solo.
 
So, I contacted (icontakted?) iZotope tech support asking why, only to receive a very unhelpful answer (their replies are usually very helpful, but this one wasn't).
 
 
Thank you for contacting us!
 
iZotope Radius allows for independent control over timing and pitch of any audio material while preserving the character and acoustic space of the original recording. Using an advanced hybrid method, it is able to preserve transients and sustained material simultaneously, making it extremely useful for modifying fully mixed performances as well as individual instruments, voices, or other audio sources.
 
If you have any additional questions, please do not hesitate to let me know. Thanks!
 

As for choosing "Groove Clip" for both the offline and online algorithms, it produced unusable results (although it introduced no artifacts IIRC, I could still hear the wavy background noise in the bounced clip). The Radius Mix algorithm introduced no artifacts either, but the tone of the voice changed as if there's a phase issue in it.
 
As I said before, both Radius Solo and Radius Solo Vocal are very good in terms of sound quality, but since BOTH of them introduce artifacts (interestingly, at different points of the song), I think the best practice is to use both algorithms on two duplicated vocal clips respectively and combine the best parts of the two rendered clips.
 
I wish the reference guide didn't just describe the Radius Solo Vocal algorithm as "Best choice when bouncing to track with audio clips of a solo voice." It should explain why it's better than Radius Solo, IF it's indeed better.

True. But if the algorithms are supposed to work better on shorter clips, I think the manual should kindly state that information for new users (so that they won't have to go through the same steps I took).
 
Also, even if it worked better on shorter clips, I'd probably prefer to use the method I mentioned above, since I can never know how short the clips should be until I've tried several times, and also strip-stretching those shorter clips one by one (it can't be done at once, unfortunately) is cumbersome.