There really is no way to answer this with any certainty. Even if we restrict our casualty list only the Mexica of Tenochtitlan and Tlatelolco, and do not include surrounding cities and groups involved in the fighting or the 1520 epidemic, we simply do not have anything resembling reliable data on this point.

One major problem is that our sources are overwhelmingly Hispano-centric and inescapably aimed at providing a narrative, not an accounting. What Nahuatl sources we do have were written after the fact and were aimed at telling the larger story, not an accountant-like focus on numbers. While we do have some of that dry quantification on the Spanish side, the fact that the accounts of Cortés and Díaz de Castillo mention that 4 or 6 Spaniards died is more of a factor of the tiny number of Spanish present. It is not infrequent that a Spanish death or capture includes the actual person's name and sometimes even a little biographic tidbit. It adds to the tension of the narrative and our tracking of the number of Spanish present, but does little to address the numbers of Mesoamerican casualties on either side. The problem of the "invisible Indian" is an unavoidable one when dealing with the Spanish sources. You'll get passages that describe in detail the number of Spanish dead, or horses killed, and then concludes with something like "and our allies also suffered greatly."

Moreover, the Spanish had little way of knowing exactly how many casualties they were actually inflicting on the Mexica and, again, we do not have comparable sources from the Mexica side that we could use for counting bodies, as it were. Even when we do have some numbers in the Spanish accounts, they tend to be generalizations (e.g., "500" "a thousand"), not a complete summation of the battle, and/or not consistent between sources.

For an example of all of this, let's look at 3 different accounts of the fighting that took place in Tenochtitlan as the Spanish unsuccessfully tried to break out of their besiegement in the city following the Toxcatl Massacre. The conflicts that took place prior to La Noche Triste, in other words.

Here is a passage from Gomara's Cortés: The Life of the Conqueror trans. Simpson:

At daybreak, the Indians attacked more violently and in greater force; so close-packed were they, indeed, that the gunners fired their pieces without aiming. Crossbows, arquebuses, and falconets, fired without interruption, made no impression. A shot would fell ten, fifteen, or even twenty Indians, but the gaps were immediately filled and the Indians seemed to have suffered no damage whatesoever. (p. 211)

So how many volleys were fired? What can we actually glean from this passage with regards to actual numbers wounded or killed?

Now here's the same incident as recounted in Díaz de Castillo's True History of the Conquest of New Spain trans. D. Carrasco:

We noted their tenacity in fighting, but I declare that I do not know how to describe it, for neither cannon nor muskets nor crossbows availed, nor hand-to-hand fighting, nor killing thirty or forty of them every time we charged, for they still fought in as close ranks and with more energy than in the beginning. (p. 217)

So was it 10-20 or 30-40? Are Cortés and Díaz del Castillo describing the same events even as we know they are describing the same time period in the narrative? Were they actually counting the number falling, or simply making (very) rough estimations?

Rounding this out, here is the post-Conquest Nahuatl version, from Sahagún's General History of the Things of New Spain, Book 12 trans. Anderson and Dibble:

And then the Spaniards fortified themselves. From there the Spaniards shot at the Mexica with iron bolts and fired guns at them. And then they placed Moteucçoma in irons. And then all the warriors who had died were retrieved, taken to places, and identified. And wailing arose from the mothers and fathers, crying and weeping over them. First they took them to their various homes, and afterward they brought them to the temple courtyard and assembled them there. They burned them in a particular place called Quauhxicalco. But some were burned only at the various youths' houses.

As noted before, this is an account for which building a narrative is far more prioritized than counting corpses.

Even taking all three of these accounts together, we have nothing to give us a solid quantification of casualties.

This gets even more difficult if we were to factor in the smallpox epidemic. Our estimates for mortality from that plague has been modeled as around 30-50% of the population. But then we have to take into account that our overall population estimates for Tenochtitlan range from 150-300K and, for the whole region, between 1.2-3.0M. So what we have instead of firm numbers is best estimates based on epidemiological models which encompass a broad range of calculated population sizes.