DXOMark has fully analyzed the new E-P3's sensor (which is also the sensor in the E-PL3 and E-PM1) and, here's a shock, it's the same fucking sensor that they've been using for over two years. No doubt about it.
It does worse than the E-P2 in DXOMark's measurements, although I don't think that the differences really constitute significance. For all intents and purposes, it does identically to its technological cousins in the GF2 and E-P2.
It does significantly worse than the new Panasonic G3, managing 131 fewer ISO points in the low-light test. That is big. It does even worse in comparison to the GH1 and GH2, to which it is very close in price. It confirms everything that we've been able to glean from the images posted thus far: all of the benefits are coming from processing trickery in the camera.