Preface: this setup existed before I was onsite. I can describe to the best of my ability how it is setup, but can't necessarily answer as to why.

We have vMix 21.0.0.61 x64 Windows 10 Home, on a laptop, with a Beheringer x18 Rack-mounted mixer, via USB connection.
The studio mics all go into the mixer, which pushes that audio back out on Aux2 for headphones, Aux3 for remote guest. We have one mic that is unmuted, and turned all the way down, and then assigned to Aux2 pre-fader, and use that as an IFB. Additionally, we do zoom call-ins by using a 2nd USB soundcard on the system which is routed to a single input on the soundboard. This channel is removed from Aux3, to prevent the zoom guest from hearing their own voice. They hear everything else, IFB included. The video from zoom is pulled via a desktop capture.

Finally, to accommodate in-studio (and remote) guests to be able to hear the audio of a pre-recorded video or audio source, all inputs that I need audio from in vMix are played on BusA. BusA is then an input on the mixer, (via USB) and then the Mainout goes back to vMix- which it streams and/or records as M(ain). I have to put in a -400 delay (yes, negative) on videos with any talking, so that the audio hitting the mixer, and then coming back is synced correctly. I don't have to worry about that, if I have audio that I don't need in-studio guests to hear. I can simply map a pre-recorded segment to M, with no delay, and the recording or streaming audience is none the wiser. But if in-studio guests need to hear it, the only way I can do it with how things are setup now, is to bring audio through BusA, to mixer input, which goes Main-out, to vMix M (which isn't heard in studio)

So- with that setup out of the way, and hopefully clear as mud--- We have problems with zoom. Problems that I think can be fixed using vMix call. Bandwidth issues on zoom that I don't see in testing with vMix call. Ability to control high-quality video of MULTIPLE guests. Etc.

But when I connect to someone over vMix call, I seem to be limited on my options. If I play their audio though M, no one in studio ever hears it. It also causes a feedback problem, where the guests are all getting their own audio back 400ms after they speak.
If I play them though BusA, we can hear it in studio, but they can't hear us. Multiple guests can hear each other. If my end product was just remote guests talking to each other, with no in-studio talent, and no need to ever hear from IFB, this would kinda work.

We're on an older version of vMix. On an older laptop. I'm not sure what options I have here, but I'd love some feedback and help. We have a new system arriving soon, and will upgrade to the newest vMix at that point- which may or may not fix these issues? I might also be opening to sound engineering this a different way to fit our needs, but need some suggestions there. However, while I wait for the new hardware to show up, I am trying to get this working for a single client project that happens before delivery.

Any ideas? Thanks all.

So would I need a bus for each incoming caller?

What do you mean by "look at my pool entry?"

So... my goal here... was 6 call-ins. Perhaps up to 2 or 3 interacting with each other at the same time, and with an in-studio host.
Also- (and this could be a version limitation) I only see Master, Bus A, and Bus B. No additional buses... ?

I'm pretty sure we're pro. Just old.
But that'll be rectified soon. I've only tested with up to 6 callers, but no issue there-- until you bring our sound config into the mix.

I will try that out in the office tomorrow.
So I understand the basic idea of putting someone on their own bus, and then making that bus an input on my board, which then plays to the main... but where in vMix do I tell the vmix call to get it's feedback? (which will have to be a separate bus, so I can mix-minus... meaning each caller needs their own bus. so most callers I can do this way is 3? a/b, c/d, e/f)

In vMix 21 Buses C to G are still available. But go to audio Settings and set them as enabled or define an soundcard output.

To define the backchannel of an caller rightckick in the thumbnail and there are audio source and video source as option. So you can define which bus each caller gets back. Also with the video. If you define the 4 different video outputs with specific inputs or PGM,PRV or MV you can select 4 differently video responses (one of them will be program unless you directly send each caller via NDI and have an NDI based video facility)

Try it out.

Take a look at my channel at https://mavik.de/live
There should be some useful information with the X18 and audio in general.

Ok- I'm getting really close... but missing something, or fear it won't work the way I need it to.

So I have Bus B outputting to my X-AIR X18 on channel 11.
Then I have channel 12, unmuted, and turned all the way down. In the X-AIR-In/Out-USB Sends, I have USB 12 pulling from Bus 6.
Then I configured each channel on the mixer to push to bus 6 at 0db, EXCEPT channel 11. It's at -∞

This worked for caller #1. They could hear me, I could hear them, and they could hear IFB.

Then I brought in a 2nd caller, and put them on the same setup. I can see and hear them, they can see and hear me, they can see each other (doing a 3-way split screen) but can't hear each other. Because of the mix-minus I've setup.

ALSO- I need to route the X-AIR buses to the input return to vMix, so if I wasn't using any X-Air busses, I'd be limited to 6 vmix callers by way of only have 6 X-Air busses? (which is another problem- because I'm already using 3 busses on the X-Air)

So here's the question- Do I have to use up two mixer inputs for each caller (up to 8 supported by vmix) -- and then use 8 outgoing busses for each caller (there's only 7) - thus limiting myself to 7 maximum callers if I'm not using the bus for anything else (I am, I use it to pull audio from videos as an input on vmix) thus limiting myself to 6 maximum callers (which might be enough- just trying to wrap my head around this more)