I find some differences between HTML data from enterprise chunk files and Parsoid API:

• In Spanish Wiktionary, page libro from Parsoid API has the template "pron-graf"'s "data-mw" attribute inside a "span" element before the "table" element:

<span data-mw='{"parts":[{"template":{"target":{"wt":"pron-graf","href":"./Plantilla:pron-graf"},"params":{},"i":0}}]}'></span><table class="pron-graf">

But in enterprise API HTML data, "data-mw" attribute is in the "table" element:

<table class="pron-graf" data-mw='{"parts":[{"template":{"target":{"wt":"pron-graf","href":"./Plantilla:pron-graf"},"params":{},"i":0}}]}'>

• In Chinese Wiktionary, page 大家 from Parsoid API could display text in a the Language conversion block:

<span typeof="mw:LanguageVariant" data-mw-variant='{"disabled":{"t":"大家"}}'>大家</span>

In enterprise API HTML data, it's:

<span typeof="mw:LanguageVariant" data-mw-variant='{"disabled":{"t":"大家"}}'></span>

• And about math formulae, like this page binomial series uses MathML now and doesn't have SVG image but enterprise API still have the SVG image. Is this related to the deprecated Math API and will the SVG images still available in the next month's files?

I hope someone could clarify are these differences intentional and whether the enterprise HTML will change in the future to be the same as Parsoid API.