TEVA - In looking for characterization data about Teva's product I can find only one instance where they have talked about their own process/product (I was hoping to find data linking it to Italfarmaco - but didn't). The response to the Citizen's Petition - but within that letter there is, as others have noted, some interesting stuff.
And what about the results for chains with mass greater than 3600 - which for Lovenox is more than 50% of the mixture? Either it doesn't have the same variability as Lovenox (which would be damning) or they didn't know at the time of their letter (see next point).
I.e. at the time of this letter they hadn't characterized the drug fully. This would, I believe, be in contrast to MNTA's version which required characterization to create the generic version of the drug. The chance of the characterization returning the right results when it wasn't 'built-in' seem slight.