An active filter is an op-amp plus resistors and capacitors, and the op-amp's own bandwidth sets the ceiling — which puts the useful range in the audio band and a little beyond, option 4.

What limits the top end. An op-amp's open-loop gain falls at 20 dB per decade above a very low corner, so the product of gain and bandwidth is a constant:

\(GBW=A_{CL}\times BW\)

A general-purpose device such as the 741 has \(GBW\approx1\ \text{MHz}\), so a stage with a gain of 10 has only 100 kHz of bandwidth. Once the filter's own corner frequency approaches the amplifier's, the op-amp no longer has surplus loop gain to enforce the transfer function, and the response departs from the design. Slew rate imposes a second, separate limit on large signals:

\(f_{max}=\dfrac{SR}{2\pi V_{p}}\)

Why active filters dominate the low end instead. At low frequencies a passive LC filter needs an impractically large inductor, since

\(f=\dfrac{1}{2\pi\sqrt{LC}}\)

A 100 Hz filter would need henries of inductance — bulky, lossy, and prone to picking up hum. The active filter dispenses with inductors entirely, synthesising the necessary complex poles with an amplifier, resistors and capacitors. That is exactly why it took over the audio range.

Why the other options fail. "At all frequencies" ignores the gain-bandwidth ceiling; "at high frequencies" is precisely where active filters give way to LC and to distributed structures; and "at zero frequency" describes a DC response, not a range of operation — though it is worth noting that active filters do work down to DC, which passive capacitively-coupled filters cannot.

Flagged because the option wording is loose, but the audio band is where the active filter's advantages are decisive and its limitations do not bite.