Well I should be clear that my experience is largely with the Peak, but I played the summit enough pre-COVID restrictions to ascertain there wasn't any particular difference in the aspects which bothered me - at which point I was no longer interested in either form of the instrument.
I feel like the Summit/Peak oscillators are not at all like the P12s, which have a more organic charm to them. As I am myself trained in DSP and implement it professionally, my general sense is that imperfections *of specific types* are essential to a pleasing, "musical" quality of tone and other types of imperfections, or the absence of them entirely, create a more alien sense - at best merely bland or boring, at worst actively painful to some sensitive to such things. Nearly all of the sounds of nature are a combination of resonance, filtering, delay, and the interactions of these upon each other at audio rate or lower modulation frequencies. For instance, in a bell the multiple resonances of the structure as they vibrate interact with each other, and as the physical structure thus physically moves to create the oscillations, the resonant frequency of its key dimensions also changes - this is a resonant form of FM in a natural expression. We become thus accustomed to the abilities of natural materials to do this work, and nature tends to work with specific ratios, harmonicities, etc. which then communicate on a psychological level to our brain with various forms of meaning and emotion. Lacking such a connection we do not sense the same things, and can often find them "unnatural" in a not-pleasing manner. I am not speaking of noise - noise is quite natural in many instances, so is distortion, waveshaping, etc.
I say all of this because one of the things I have noticed about the ongoing "perfection" of digital oscillators and even digital filters (which I personally find very disinteresting - please do note that I don't mean good and beautiful music cannot be made from them - I have personally made "acoustic" and very pleasing timbres with my various Nord Leads) is that the more they strive for high bitrates and the avoidance of antialiasing and other properties via brute force (e.g. sample at such a high rate that the aliasing becomes inaudible, then either filter it out or don't care at all about it) -- the more pure the oscillator sounds at first, but also the more boring it is without a significant amount of additional work to alter that perfection which was attained at such efforts to begin with. The opposite approach - designing a normal nyquist-rate or low-oversampling ratio (e.g. 2x or so) oscillator that doesn't care much about antialiasing - just sounds poor to begin with, but that is an obvious and unnatural variant and thus is more readily noticed and rejected. I don't have a perfect answer (there is none in the discrete-time world, I don't believe) but oscillators which run at roughly 48-96kHz and use careful mathematics rather than brute force for aliasing reduction have consistently caught my ear as "musical" in my estimation - though this is not a hard and fast rule, just something I notice in my own auditioning.
I suspect, only through a preponderance of experience, that this may be related to the fact that very high harmonics are often found in nature with quite a lot of complex variance (audio rate FM, I would suppose) when they do occur - not necessarily enough to be obvious, but enough that the additional sidebands caused by the variance are subtlely picked up. Nature is full of extremely subtle modulations like this and when they are not there we miss them. One of the issues, then, with extremely high bitrate oscillators is that in many cases they process their modulation at this high rate as well and often that means that artifacts caused by "lower fidelity" oscillator architectures are gone - and my core theory is that these artifacts of various kinds are a substitute for the complexity the ear expects to hear in the modulation of the fundamental timbre. In other words, rather than getting a pitch modulated at near-audio rate in steps (which cause artifacts on top of the fundamental and the modulating waveform), you get a pitch perfectly modulated in beyond-audio-rate which yields ONLY the exact (in the audio spectrum, at least) sidebands of the modulation, nothing more. This nothing more is the problem, not a feature, in my estimation - though again I must be quick to point out that any artifacting that might "solve" the problem is only a substitute for a natural complexity and not ideal. The artifacts can be of various natural or unnatural aspect and of course not all of them are pleasant or "musical".
Anyways, enough of the bloviating. You can see my point, I trust, that perfection in digital oscillators is not, in and of itself, any benefit without immense complexity in modulation, and that makes programming a great and natural feeling sound much more difficult and a process of greater effort (and in many ways undoes some of the point of having such oscillators in the first place) - and that for synths which have artifacts and other imperfections which *do* lean musical, often those can be preferable and a faster way to the core timbres that one might be looking for.
The Peak/Summit, to me, are exactly this: bland by nature, and while you *can* work hard to make them sound interesting (and, I won't say I've ever heard it, but I'll allow perhaps even "glorious" *might* be possible), there is nothing in them that makes them remotely worth that level of effort or investment for me, when I can get that and much more in spades in many other instruments (I even prefer the quick sounds I get from my Nord Lead 3, frankly, for at least some subset of the possible sounds I might want to make - though I'll give it to the Peak for strings over the Lead 3 - but I'd not choose either of them for strings, for instance, to begin with).
Onboard effects on a synth are not particularly valuable to me (although an onboard delay or 4 is most welcome, because I view delay as fundamental to a timbre and not an "effect" that is added onto it later), so the Peak/Summit's value there is of no interest to me. And their filters are certainly nothing to write home about - they are pleasant, competent filters. But again, lacking anything special.
Overall, while I think I am generally leaning towards the digital oscillator/analogue filter as a sort of ideal architecture for the stuff I find myself most appreciating, I don't find the Peak/Summit to offer a compelling argument.
Take all this with a large grain of salt or a glass of seawater as you prefer and know that it is one opinion and not gospel, as it were.