There is no one answer. It's not a deterministic value. There's manufacturing variation plus it depends on degradation, of course, but also usage conditions. A battery that is storing "75 kWh" of energy will deliver different amounts based on it's temperature and discharge rate.
The EPA test data is the most reliable battery capacity estimates I've seen for Tesla. This data is based on the amount of energy the battery can deliver from full to near shutdown, independent of a "0% or 0 mi range" indication to the driver. Tesla includes about 4.5% (or ~3.4 kWh) buffer at the bottom of the battery below 0% indicated SOC. This is included in the usable battery capacity estimate, but should not be relied upon while driving. It's there to ensure, due to all the challenges associated with estimating remaining battery energy, that the car doesn't need to shutdown
above 0%, which would lead to very disappointed drivers.
Below are all the published Model 3 and Y Long Range battery capacity estimates from the EPA testing. All of these batteries were evaluated using the same measurement technique (SAE J1634) and even I think the same dynamometer at their Kato St facility. The J1634 standard requires cars be aged to at least 4,000 mi, therefore these batteries have already experienced a portion of the initial degradation process common to all Li-ion batteries.
It's clear there is variation that is likely attributable to cell manufacturing tolerances, or potentially how the cars were "aged" prior to the test. The
average of these 8 samples is 78.6 kWh. Subtracting the lower 3.4 kWh buffer would yield an average
75.2 kWh available from 100% to 0% indicated SOC, but the range for these samples is 73.1 to 76.4 kWh.
View attachment 558947
I'll also mention that, to date, there are have been two iterations of the 2170 battery cell used in the Model 3. I don't know what the changes include, nor whether they affect capacity. I suspect the 2nd iteration was in use before the 2020 model year cars started being manufactured and all Model Y vehicles have the 2nd iteration, but have no evidence.
Links to the EPA Applications from which this data was gathered:
https://iaspub.epa.gov/otaqpub/display_file.jsp?docid=48305&flag=1
https://iaspub.epa.gov/otaqpub/display_file.jsp?docid=46585&flag=1
https://iaspub.epa.gov/otaqpub/display_file.jsp?docid=48712&flag=1
https://iaspub.epa.gov/otaqpub/display_file.jsp?docid=49398&flag=1