I've been reading Thomas Asbridge's The First Crusade and in it he repeatedly states that the numbers given by the Crusaders of the Muslim armies are likely to be smaller in reality due to the difficulty of actually counting how many there are and the chroniclers exaggerated the numbers to make the victories more heroic and the defeats into tragic fights against overwhelming odds. Added with the general consensus of Herodotus hugely exaggerating things like the Persian numbers can the size of armies in any battle where we only have accounts from one side in the medieval and ancient world actually be trusted?
Try the modern world. Are there Russian's fighting in Ukraine or aren't there? If there are, are there 1000 regular troops, or 1600 regular troops? deception/bluff regarding military strength has always been a part of telling stories about a communities strength, past or present, the best we can do is determine what the purpose of the history/story was in order to determine what types of biases were/are being applied to the figures. In other words, even having records from both sides might not clarify anything.
In addition to this, there are a lot of other factors that can mess with giving numbers. For example, there are arguments within biblical studies over what base system is being used at times, or whether we are counting thousands of individuals, or simply counting family units.
In other words, the "correct" number can be very opaque, however this is not necessarily a total loss. While knowing the economic or social strength of a political entity based on the amount of armed men it can field is interesting knowledge, this type of a number is often not the window to the past it might seem, in fact such numbers can very easily be plugged into our own pre-existing narratives of interpretation and allow us to ignore many far more revealing bits of information.
Imagine a hypothetical inscription where we read that 10,000 soldiers of one group (group A) allied with 10,000 men from another group (group B) to stand off group C (who are listed as a scary 50,000 strong!). The chances of the two groups being equal is fairly low, but if you start looking for reasons why the chronicler might have felt it important to show apparent equality you may learn important knowledge about the social connections between these two groups. A similar case to this is that of the Medes and Persians; despite not being proportional in size or contribution, in official artwork of the period there is an almost even distribution of Medan and Persian style clothing hats and beards. The lie gives a stronger message, and more understanding of the people than the truth does or would have.