The different verse counts reflect differences in where early reciters understood one verse to end and another to begin, not differences in the Quran's wording. They establish variation in verse numbering, but they do not establish that regional communities possessed different Qurans.
Why the figures genuinely differ
The objection is fairly stated: the familiar figure of 6,236 is the Kufan count associated with the reading printed in most Hafs copies. Other traditional systems give figures such as 6,217 or 6,214 for the Medinan counts, 6,210 for the Meccan count, 6,204 for the Basran count, and about 6,226 for the Syrian count. Modern editions associated with Warsh may therefore display different verse numbers. If "flawless transmission" meant that every community must preserve one universally fixed set of verse divisions, this would be a real difficulty.
But that is not what the claim of Quranic preservation means. The disagreement concerns add al-ayat, the counting of verses, not the words recited between the beginning and end of the surahs. A verse marker is a later written sign indicating a division in the recited text. Early manuscripts did not all use the numbered verse symbols familiar in printed Qurans.
One text, different stopping conventions
The early scholars transmitted more than one regional system for identifying verse endings. These systems were based on reports about the Prophet's (PBUH) recitation and on the practice of the Companions and their students. A reciter might stop at a particular phrase in one transmitted system, while another recognized that phrase as connected to what follows. Questions such as whether a particular opening formula is counted separately, and where certain passages end, can alter the total without adding or removing any words.
This is why classical specialists treated verse counting as a distinct discipline. Works on rasm and add al-ayat discuss these regional traditions precisely because the underlying consonantal text and recitation remain substantially the same. A person comparing Hafs and Warsh will encounter differences in reading in some places, as part of the recognized canonical readings, but the different printed numbers are not evidence of missing passages.
Indeed, We have sent down the Reminder, and indeed We will preserve it.
Quran 15:9
What this argument does and does not establish
It establishes that Muslims historically preserved more than one accepted method of marking verse boundaries. It also shows that "the Quran has exactly 6,236 verses" is not a universal theological claim; it is a statement about one counting tradition.
It does not establish that the Quran contains different regional contents, or that its wording cannot be traced through the memorized and written transmission of the community. To turn a difference in numbering into evidence of textual corruption, one must first assume that divine preservation requires one mandatory numbering scheme. The Quran itself does not make that claim, and the classical tradition never confused verse counting with the preservation of the revelation's wording.