If the dataset is assumed valid and you avoid multi-word 'make' fields, LanX's approach would certainly seem to do the trick with this dataset. Note that  $extended_model has to be 'fixed' if the field does not exist, otherwise it's undefined.

>perl -wMstrict -le
"my @records = (
   '2011 Chevy Camaro',       '2011 Dodge Ram Crew Cab Short Bed',
   '2011 Ford F150 Platinum', '2011 GMC Cargo Van Extended',
   );
 ;;
 for my $record (@records) {
   my ($year, $make, $model, $extended_model) =
     split ' ', $record, 4;
   $extended_model //= '';
   print qq{'$year' '$make' '$model' '$extended_model'};
   }
"
'2011' 'Chevy' 'Camaro' ''
'2011' 'Dodge' 'Ram' 'Crew Cab Short Bed'
'2011' 'Ford' 'F150' 'Platinum'
'2011' 'GMC' 'Cargo' 'Van Extended'
[download]