As Nikon moves more and more into shutterless (or at least electronic shutter) cameras, read-out speed of the chip becomes important, as it will define how well motion is captured without a rolling effect.
It's important to note that mechanical shutter speeds faster than 1/250 (on most cameras) also "roll" the subject, as the shutter at that point becomes a moving slit. Thus, an electronic read-out speed of 1/250 or faster means that an electronic shutter is going to be no worse than a mechanical shutter (though there is a slight difference in rendering).
Here are measured full sensor read-out speeds for electronic (silent) shutter on the Nikon Z models for still photography (all these values are for raw files; JPEG pre-release capture can be different, as it uses video modes):
- Z50 — 24.6ms (1/41)
- Z5 — 100ms (1/10)
- Z6, Z6II — 50.8ms (1/20)
- Z6III — 14.4ms (1/69)
- Z7 — 66.7ms (1/15)
- Z7II — 65.6ms (1/15)
- Z8 — 3.7ms (1/270)
- Z9 — 3.7ms (1/270)
- Zf — 50.5ms (1/18)
I mentioned "still photography." In video modes—and remember that Pre-release capture uses a video mode—the full sensor readout can be different. On a Z8 and Z9, the video modes are slower, on a Zf, Z6, Z6II, Z6III, Z7, and Z7II the video modes are faster.
We can argue over what constitutes enough electronic rolling shutter to be visible, but this tends to be somewhat dependent upon subject and speed. The classic test is to try to freeze helicopter blades in rotation. Little to no impact (straight blades) is the goal, but at slower effective shutter speeds, the blades will start to appear curved. The Z8 and Z9 arguably do quite well with this test, effectively giving you the same result as you'd have with a mechanical shutter. The second best camera in this regard is the Z6III, with the Z50 being the surprising third best. Pretty much all the remaining Z cameras I list above have too much rolling shutter impact to render fast motion without visible artifacts.