This is a really bad model as it makes money part of an equation which should only be governed by scientific concerns: should this paper be published or not?
Now, researchers is poor countries may be able to access existing research but they won't be able to publish their findings…
And even thought it kind of solve the problem of mass access to scientific publications, it doesn't deal with the fact that a lot of public money is going in the pockets of private academic publishers for no good reasons. With this model, research is still paid for at least 3 times (for doing it, reviewing it, and now publishing it instead of accessing it) by universities and research institutions, while publishers are making an awful lot money for hosting PDFs. This money which could be used to do more research, as only a tiny part of it would be necessary to support the necessary arXiv-like infrastructure.