In my application I would like to use it to drive a high impedance input, and the pull up current for the open drain is significant compared to the rest of my circuit.
There is no push-pull version of the ADCMP380 and no exact replacement option as well. But you may want to check these push-pull comparators also by Analog Devices: ADCMP356, ADCMP371, ADCMP380.
ADCMP356 has 0.6V on-chip reference and has non-inverting input. But if you require a specific reference voltage, then the general purpose ADCMP371 and ADCMP380 may do, in which you have to use external reference. ADCMP380 has shutdown pin if you also require such.
Is the pull up current really a problem? Consider with normal operation, the duration of the reset should be short and only happens very occasionally?
also the ADCMP380's leakage current is 5nA max over temperature at worst input voltage, the typical value should be much less. since you are driving a highZ input, you could use a very large pullup resistor to reduce the pullup current.
I can't help you with the push pull situation, but I may have a solution to make your divider dream come true.
take a look at ADM8611.being voltage supervisors they operates slightly different to compactors in that they have a build in delay. but i don't see that being a issue for your application.
note not all threshold options are available in production, in fact only 2.63V is currently available for ADM8611, there's a also 2.32V on ADM8613. they may work with your 3V battery.
It's not a completely unacceptable problem, but the pull up current is a large and unnecessary part of the budget. I'm using the vishay SiP32431DN load switch with the ADCMP380 or other comparator to implement a low current, low battery shutoff so the pullup current would be in play in the battery discharged state - which may not be a short event, and removes all other load currents from the system. The vishay load switch has an Iq of pA, and the attraction to the ADCMP380 is it's low 92nA Iq. Even a 10M ohm pullup will require 300nA at 3V, which is about 3x the active devices of the circuit... Admittedly, the sampling divider is going to be on the same order as the pullup, but it is also more necessary to the operation of the circuit.
I think my question is answered - no push pull ADCMP380 is available. If ADI is listening - count this as my vote for a push-pull ADCMP380 option! While I'm dreaming, eliminating the need/impact of the divider would be nice too. Perhaps a sampled time comparator, that only briefly connects the divider at regular intervals to check the comparator voltage?