The variables needed for machine learning are different than a standard air quality model. It's actually a much shorter list because you don't have to numerically model atmospheric motion/chemistry/deposition.
Most AQMs have a spatial domain that has horizontal and vertical depth. However, the machine learning forecasts I've seen are only for point ...
From Seigneur C., Dennis R. (2011) (free access here, probably not in the final published form):
Inputs to air quality models include the emission rates of primary air
pollutants and precursors of secondary air pollutants, meteorology
(three-dimensional fields of winds, turbulence, temperature,
pressure, boundary layer height, relative humidity, clouds and ...