None of them intentionally give the live virus to their subjects.

Challenge studies get performed on non-human (animal) test subjects (this was done for the Biontech/Pfizer vaccine), and there's a volunteer human challenge study planned in the UK for January. A challenge study is one where you intentionally give a live virus to your subjects.

Arguably, releasing your subjects into an environment full of live virus, such as the United States, is a kind of challenge study, which rapidly delivers efficacy data. This is why approval could happen so much faster than usual for these vaccines.

That would speed up the test, sure. But as far as the quality of the results goes, why is that different from the existing test where a large number of vaccinated (real and placebo) subjects gets exposed/infected? It's still testing efficacy.