It depends on what you mean by a finite distance. If they are farther apart than 2D2/wavelength, where D is the largest width of an antenna, or possibly the sum of the largest widths of each antenna, and you use the shortest wavelength, then you can get a good answer using the far-field source. This is equivalent to using the Friis equation with far-field gains.

If they are closer than this you may need to generate a near-field source from one antenna to illuminate the other, although you may get useable results even at D2/wavelength. I assume that are not so close together that they are in each other's reactive near-field.

Thank you very much! Yes, indeed, when I wrote about the finite distance, I meant the intermediate near zone. The thing is that the description of this type of sources CST states that they can be used to estimate the mutual coupling between antennas. However, all the papers found on the use of such an approximation consider omnidirectional monopole antennas located on a common platform and, of course, the distance between these antennas corresponds to the far zone criterion. This raised doubts about the applicability of such farfield sources for antennas with a large aperture located in the near zone. Thanks again for the clarification.

I think that you can use field source monitor data (FSM files) to use the near-field of one antenna to illuminate the other as a near-field source. I haven't done this but it is described in the CST help. It is the way I would expect to try. It won't give the change in impedance of the illuminating antenna but that may not be significant. Alternatively MLFMM methods should give everything. Again, I haven't used them.

Thank you, Malcolm. I'll start with near-field sources. As it turned out, the task of assessing mutual coupling took longer than expected.