Sure, BJTs are quite capable in this range. Audio amps can easily go higher in both frequency and voltage. You do need a strong enough driver; additional transistors are usually required, say in a biased Darlington configuration (or a full discrete circuit, because audio is just like that..).
Note that, if your outline conditions are true, then maximum power is dissipated around half load, 245W. You'll need a couple transistors in parallel for that, and a fair sized heatsink, give or take if it's continuous or intermittent. If the load can take any impedance however, the worst case is a short circuit and the maximum dissipation is four times that or just shy of 1kW, a much bigger challenge!
Note also that, for 50kHz at that current, and reasonable drive voltages (say <10% of total supply), you need less than 0.25uH in the output ground return path. This suggests noninductive resistors and short wiring, though obsessively short shouldn't be required. (This is equivalent to about 20cm of loose hookup wire.)
Likewise, the maximum load inductance must be below 2.5uH, else the output will be clipping.
Don't forget some kind of transient overvoltage protection. This can be as simple as a plain old TVS across the output, or a zener from input (gate/base) to output, etc. (Probably with a ferrite bead so it doesn't cause oscillation.)
MOSFETs are easy to parallel if you have some voltage to spare: just put a relatively large source resistor under each, such that Vs > ΔVgs(th). The same is true of BJTs, but BJTs have a smaller ΔVbe between each other, so Re is just about inconsequential. For MOSFETs, this is reasonable practice at high voltages; at this voltage, it might be better to use independent servos (one amp per transistor) and only wire the outputs in parallel. The price is saturation voltage, in turn requiring higher supply voltage for the same load range, and bigger Rs's (more watts).
Tim