If you believe this paper from last year, roughly one in five current peer-reviewed articles is now to be found open-access. (I'm not sure why the paper is only getting press now, but the world is weird.)
Numbers are bunk, let's just say that up-front. I don't believe these numbers; I don't believe most numbers around open access. It's too hard to draw a circle around what we're trying to find out. This study (as with most in this area) omits an important fine point, as well: how much of this open access is, y'know, legal? The authors found papers "on the home pages of the authors." Let me tell you, researchers don't know from copyright, much less publishers' policies surrounding same. I'd be pleasantly shocked if half those postings were legal.
Still. Twenty percent. One in five. It's a long way from zero.
Some days I feel that what I do is stunningly pointless. This… is not one of those days.