I just went back and looked at my email correspondence with Schoeps, because this all rang a bell with me. In January, 2008 I noticed the same discrepancy and raised a question about it.
Apparently an internal misunderstanding had led to the release of some projected specifications, while my contact there (the engineer in charge of their publications) was apparently under the impression that he'd been given actual, measured results. (Those came only a little later.) It was a mistake, in other words. The specifications on the Web site now should be the correct ones. By that I mean the ones on the HTML Web pages themselves, not in any downloadable PDFs.
However, the sensitivity was never intended to be the same as that which you would get with a CMC 6-- amplifier. The email correspondence reveals that another mistake occurred: My friend had clipboarded the CMC 6-- specifications into the new material for the CMR as a starting point, and at one point had simply failed to overwrite the CMC 6's sensitivity specification with the CMR's. So, Murphy strikes again.
Please note that the CMR circuit is optimized for minimum power consumption, so the dynamic range shouldn't be compared with that of the CMC 6-- amplifier in the expectation that they'll line up exactly. You can't get something for nothing in physics; you can get better dynamic range when you aren't on a super-tight current budget in an amplifier circuit. Thus the equivalent noise of the (much lower-current) CMR circuit is a little higher than that of the CMC 6-- amplifier, and not only as a function of the capsule polarization voltage.
--best regards