in any test i’ve ran, vst plugins always render 1:1*, and pass the null test. unless of course there is some random element explicitly introduced in the code of the given plugin, so as to approximate some typical behaviour of a hardware box it’s trying to imitate, and so on.
*this however only applies when running the test using the same buffer-size of your audio interface. things will often sound different on higher buffer sizes, and rapid automation changes especially will impact the resulting signal, as the automation becomes ‘slower’ with higher buffer-sizes.