For continuous operation using a superconducting setup may be the best option if issues of insulation (thermal) can be dealt with appropriately. I have seen an aero-space setup based on an internal 25cm dia space, 25cm long, which needs 198,943 ampere turns to generate 1T in the 25cm length (the return magnetic path was created from 25cm dia iron sections, cut with 45 deg ends to allow the creation of a return path so that the net magnetic path length was that inside the solenoid arrangement). 8 coils using Cu foil were used, each being ~30mm wide and fairly tall, with an ID sufficient to fit over the 25cm dia of the required space, and the total width of the 8 wdgs being ~250mm, there were little wooden spacers or pegs in the wdgs every so often to allow oil circulation without raising the total wdg dia too much. Each winding was of 2486 turns, and of 10 ohms at 20degC, each requiring 10 amps and dissipating 1kW each, i.e. ~8kW in total, the whole being immersed in a light oil bath with oil stirrers and cold element radiators (heat absorbers) in the oil bath. 8 power supplies (constant current, 10A, 150V max, gate-able) were connected to the eight windings to provide constant current and hence a fairly constant field in the space. This worked reasonably well for fairly high duty cycles (50%, 1 minute on, 1 minute off) and the power supplies could be replaced with AC sources for AC fields and later on much higher pulses were used to generate high peak fields for special testing (with and without the external iron magnetic path) , much quicker and cheaper to build than cryogenic, and can be used in air for short pulses.
I note the comment below
As a numerical example, you need 800 kA/m to generate 1 T by an air coil, respectively a current of 8000 A through a coil with 1 winding per cm.
Not sure where the "per cm" comes from as if you wind 800,000 turns on a toroidal shaped former, air cored (of a suitable size) with an average path length of 1m you will get an average field of 1T internal to the toroidal, with 1 amp in the winding. Similarly if you arrange a single turn on the same toroid and get even current flow (800kA) in the one turn there will be 1T (ave) generated in the 1m path length of the toroid. Very similar results are obtained for a straight solenoid structure purely in air where the return path is so voluminous its effective length can be considered quite short. (I think the total mag path length is then approx twice the solenoid-structure length).
Regards, Orson Cart