232nm is a bit funny value, because at 45nm the deviation can be 5-10nm from the litography as I know, I am not sure, but ok. If you can use smaller size probably it could help on a lot of things by the way.
You used 15.5nH ideal inductor. Chip inductor is absolutely not ideal, has a lots of parasitics, and takes huge area, recommended to use it over 1-2GHz, but you didn't mention your input frequencies.
If you use an LNA before this mixer maybe you don't need inductive degeneration, on silicon the matching criteria between the LNA and mixer is not so critical if the distance is small enough between them.
I think switching transistors can operate in triode region, it is not a problem, and with the 500uA current the value and area of them can be low enough. For direct conversion probably this the most suitable way... what you also forgot to mention, how is your output (intermediate frequency, bandwidth).
And you didn't share the noise figure, IP3 or P1dB value either. Very important specifications.