It appears the ratio holds. A user reporting 48 Gb available to redshift ( 6GB up from 42 GB if I remember correctly ) on his m1 max. I have a m1 max MBP. Will test myself later this week.
Really depends on the applications being used. Some applications allocate more buffer from memory if they see the system have more to offer, thus reducing the "VRAM" for the GPU.
If we just look at memory consumed when macOS boots up, we will likely see a plateauing of memory used if plotting memory used against total memory installed. The larger the amount of memory installed, the more memory is required by the OS to keep track of all the memory pages, but as the amount of memory grows, it will outpace the memory used, without any apps asking for memory.
Most OSes will also reserve some memory for essential services like networking, file I/O, etc, so the more memory that's installed, they more will be reserved, but I would think it will also plateau off after a certain size. But if there's more memory not use when a file is loaded, most OSes will just use the free memory up to cache the files loaded, but these allocated memory regions will be freed up if applications ask for memory if there're no more free regions available.
In short, I don't think there's any magic ratio.
