While nuclear thermal is better than straight chemical it still fairly fuel inefficient when compared to electric designs, (ion, Hall effect, VASIMR).
- NERVA - 850 Isp
- ION - 1700-4100 Isp Average
- Hall Effect - 1500-3000 Isp Average
- VASIMR - 3000-30000 Isp (tuneable in real-time)
So while all of the above require a large amount of power the electric ion/plasma propulsion gets much better specific impulse (mileage). With ISS assembly techniques and a heavy launch vehicle like SLS or Dragon Heavy we can build a vehicle of whatever arbitrary size you (and your budget) want without a nuclear thermal booster. For interplanetary flight nuclear electric works better because it requires less propellant for a given delta V than even nuclear thermal.
It also means you don't have the nasty nuclear byproducts in the reactor until it's already safely on orbit because the reactor isn't required until it's on orbit.
EM drive is interesting. There's an OT thread I just updated a few days ago about the paper in peer review. I actually have a friend who was part of the group investigating it for a while. They hypothetically think it's pushing off quantum vacuum virtual particles that pop in and out of existence, (see Casmir force).
It's still not clear the small amount of anomalous thrust isn't caused by experimental error. The exciting thing is as they reduce experimental error the thrust remains. By analysis an EM drive would have an equivalent specific impulse of over 10,000,000. It would be a huge breakthrough if it actually does check out.
They are also puting the EM drive in the path of a laser interferometer to see if it works by warping space.