Can you program them in parallel with the same control tags but different source, mask, and destination data areas? That was a trick used in the PLC5 days...
You can examine the done bits from each in series to limit sequence advance in a normal sequencer if you need a "wider" sequencer input size (not having seen this AOI), but I like to assign them to an internal bit if I use the SQI or MEQ with indirect addressing, so that when there is a mismatch when monitoring them online, it is easy to go to the individual word or DINT in this case, right away.
But I always add PLC logic to generate HMI code that will report the names of mismatched sequencer inputs step by step so you need not troubleshoot from RSLogix when something halts or delays the sequencer. You can store the timing of each change of state of the inputs in a trace table just by recording the timestamp at every change of state of the inputs in a separate array with a copy of the input data. That info can be used to nail down exactly which input device is slowing the process when it is neck and neck between multiple culprits, and a few milliseconds improvement can be valuable.
I normally use MVM and MEQ instead of the RSLogix varieties because I have had a need to otherwise alter the step numbers, the SQO outputs become unconditional and the position altered with a move command after a normal automatic branch from the SQI logic or from the operator controls when they need to back up and repeat a few steps or skip over one or more to get to a spot from which they could repair or remove misapplied components.
I did not like the way the SQI and SQO instructions affect the position in the classic example usage, and had trouble getting a SQI to "fire" when it entered a step in which all the conditions were met, it should advance again, but in a PLC5 and SLC as I recall, it required a false to true transition and would hang under certain material length settings in my particular case. I patched that up with a parallel MEQ instruction, then just erased the SQI from the code and and left the MEQ.
I still like the SQO for the RSLogix viewing capabilities, and the built in DN bit, but a MVM with indirect addressing is possibly more efficient and does the same thing with the data fields. There is nothing wrong with writing to the POS of the SQO with a MOV and putting the SQO instructions on unconditional rungs. I never drive real outputs with the destinations of the MVM or SQO, they are later examined in series with mode and basic machine state to drive the outputs.