I haven't used Gaussian in a while but for Avogadro i can tell you that there is a weird difference between Linux and Windows: in the 2018 Linux version, imaginary frequencies were displayed correctly while the Windows version couldn't handle them for some reason.

Jürgen Weippert Thank you very much for your answer. But then that would mean that I have not reached my minimum energy conformation yet?

The most puzzling thing for me is that these frequencies are "isolated" from all the others in the output, and have no associated xyz coordinates so they cannot even be visualized. Why is that the case? What do they actually represent?

If this imaginary frequency is displayed in another editor, then it would not be a minimum conformation.

As I said, my Gaussian experience is rather dated, but you could check whether in your regular list of frequencies there are 3N-6 sensible positive frequencies. In that case it's a minimum. Also, I just searched around, maybe the answers in these discussions will help you:

https://www.researchgate.net/post/Why_am_I_getting_low_frequencies_in_Gaussian_09

https://www.researchgate.net/post/How_to_eliminate_low_frequencies_in_Gaussian_calculations

Hallo, Cuong Dat Do!

The first 6 low frequencies in the Gaussian output correspond to 3 translational and 3 rrotational degrees of freedom. Thus they have nothing to do with your system stability.

The next 3 low frequencies (in the second line) correspond to the lowest 3 vibrations and should be positive.

Here is the corresponding part of my Gaussian output for a stable SH4(+2) geometry:

Full mass-weighted force constant matrix:

Low frequencies --- -0.0047 -0.0047 0.0018 34.4232 34.4232 34.4233

Low frequencies --- 960.6996 960.6996 960.6996

Diagonal vibrational polarizability:

3.9610226 3.9610226 3.9610226

Best wishes

mb