Having ordered 5MM, 10MM "standard" LEDs, and a 3W "Power" LEd from 1 manufacturer in the past 2 months, my observations are....
They are roughly the same output, for the same wattage consumed, at their "typical" ratings.
How does that relate to marketing? It doesn't, really....
compare a naked 700mA 3.4V "power" LED to 20mA 3.4V 5MM LED.... 700/20 = 35 standard LEDs to equal the watts of a power LED... so if you put 35 10MM LEDs into an integrating sphere and measured, and put 1 3W "power" LED into that sphere, both driven at 700mA total current at 3.4V (or 2.38 watts), the sphere would roughly see the same light total for each type of source. That may not be precisely true, but it gives an example of the general situation. In fact, the Cree XR-E uses the same die technology in it's 700ma version as the 20ma version, just designed and scaled up significantly.
How does that relate to off-the-shelf performance, etc? Not at all, marketing, measurement methods, optics/reflectors, etc all have a huge influence on the package and results. It is going to be much harder to collimate the output from 35x25-degree LEDs spread over 3sq inches of area into any kind of tight throw, than it is the output of 1 3sq=mm LED, so low-power LEDs even in large arrays are limited in "flashlight" abilities.
For usable flashlight output in a 1W plus light, a single LED package is going to be preferable from a design standpoint, all else being equal. But at the end of the day, 35x5mm at 700ma total is electrically equal and roughly output equivalent to 1x"Po-Led" at 700ma.
(In actuality, the 35x5MM LEDs probably would be BETTER than the power LED due to higher efficiency of operation, the Cree XR-E is the first "power-LED" that really scales much more linearly into the 350+ mA range (in other words, as the milliamps go up, the efficiency drops, rapidly in some cases... the XR-E is about the same efficiency at 350mA as most LEDs are at 20mA, which is a huge jump in efficiency).