Let's go extreme!
Say a 10km radius for 63km of track, and wind a 1000 kg load up to 12km/s for interplanetary launch capability, or a 1200kg load up to 8km/s for LEO.
That's really simplistic of course, because it completely ignores how our launch vehicle is supposed to cope with exiting the atmosphere at 12km/s, and it still needs to carry fuel to change from launch trajectory to orbit, but I don't know how to figure those parts out. I'll just wave my hands and call it 'minor details'

.
So anyway, that's maybe 1500 tonnes weight on a small section of our maglev track, which is going to bump up the engineering challenge. Our costs are also going to be significantly higher than base maglev, because we're running at much(!) higher speeds, which again makes engineering fun.
So, let's say $200million per km, plus padding, for a nice total of $15 billion, or 3 years of space shuttle budget.
Would it be worth it?
A
Falcon 1e can lift 1000kg to LEO for about $10million per launch.
So if this system could reduce that cost to $1 million per launch, then we can recoup costs in 1500-2000 launches.
Since there will be few launch windows each year, and the market for LEO launches isn't that great... we should break even in about 100 years. Hmm. Will definitely have to keep construction costs at around that $2 billion mark after all.