Simulating mimcap capacitor self-resonance frequency problem in gpdk045

I'm trying to simulate gpdk045 mimcap capacitor self-resonance frequency, but getting strange results (if I change the capacitance by adding additional or removing mimcap capacitors, marker frequency is not changing, only the amplitude of the slope) . Connection schematic and AC analysis results are shown below:

If I disconnect all mimcaps, and connect only nmoscap2v from gpdk045 library, with the same capacitance 152 fF, simulating results are normal:

Also, when 152 fF capacitor from analoglib is connected, results are the same, as with nmoscap2v (slightly different, but still as should be):

If I connect mimcap in RC circuit, after transient analysis, I'm getting normal results (can calculate the right capacitance of mimcap -  37,57 fF):

So by RC transient analysis I can prove that mimcap model is working as expected, capacitance at 1 Tau matches specified in schematic.

Could it be some gpdk045 mimcap model bug?