And you can get 1300-1500s ISP from a liquid core NTR, too - or 3-5000s from a gas core NTR. Sure, no one's ever built a liquid core NTR, but there have been designs made.

And needless to say, "tens of newtons" is not the projected thrust of a liquid core NTR. More like a few hundred thousand.

Scalability, and scale matter too. I believe that solar electric thrust to weight ratio is very favourable with modern solar cells, and scalable.

Arcjets can be really tiny, and you can have hundreds of them. Given that you will also have to get huge amount of electricity for crew needs, you will have to pack solar cells anyway. A bimodal NTR will be even heavier, and require even bigger vehicle to legitimise its use.

Only minimum scale really matters - past that you can just cluster engines to get the desired acceleration. Minimum scale for any sort of nuclear thermal rocket is below what you'd want for a manned interplanetary mission, and is thus not relevant.

Solar power plus ion engines is in the 1/2000 TWR range as far as I'm aware. That means millimeters per second squared acceleration of your total craft at best, and means you basically don't save any time over a standard minimum-delta-v Hohmann transfer - 0.001 m/s^2 continuous acceleration gets you 2 AU in ~400 days. It could also do 66,717,283km - the Earth-Mars distance at time of writing - in ~189 days. A Hohmann transfer from Earth to Mars is 259 days. And, of course, the above numbers don't take into consideration matching velocities or escaping Earth's gravity well in the first place. [1] does a good job of describing why the power supply is the primary limiting factor here.

Liquid-core NTRs [0] aren't bimodal, and I'm sort of confused what you'd mean by bimodal here in the first place.

0: http://www.projectrho.com/public_html/rocket/enginelist2.php...

1: http://www.projectrho.com/public_html/rocket/enginelist.php#...

Above I meant arcjets under solar-electric.

6.4kg 50N 30%-40% efficient hydrogen arcjet, and 1kg/kw solar panels will probably scale up until a 1.5-2kn, which is really a lot of for a relatively efficient, 1000s+ ISP engine made using existing material science.