The primary value of a release is the extra exposure it garners, the spike in upgrades that happen as a result, and the discovery of bugs (or maybe just user perspective) that flow from it. The cost is extra QA before release, and bug-squashing following. The followup is also a benefit, esp to the extent that reporters are told "the fix is in dev-branch, rXYZ", and they install and report back. The quality of those reports may be below those against dev-branch, given that dev-branch users are more willing to be on bleeding edge, and are presumably more skilled. OTOH, more reports (of either success or failure) should at least provide more status (given ~20 platforms, thats not a trivial info improvement).
But all that is fairly obvious, and value added is hard to quantify.
The QA is the real question: release quality is a judgement call, requiring expertise, and one thats heavily loaded compared to any particular bug hunt & squash. A page on test setups and testing is essential (I note that Orca gave a link to one, Im going to read it. Probably everyone here should).