Now Cree also realized that enthusiasts overdrive LEDs (sometimes to insane amounts) 
(And likely it also serves for the purpose of evaluating new thresholds or key selling points)
I don’t know how the impact is from LED tests here in BLF to LED and flashlight manufacturers, but it seems that at least Cree is aware of this trend. There are some interesting points of Crees testing method though:
- Cree tested the light flux in conjunction with Tsp, not only for max light flux I do.
- only 100 W thermal capacity of cooling plate (likely based on peltier elements). From my experience with peltiers, above this threshold the temperature rises sharply because the peltier has no thermal mass and is not able to maintain a low temperature any longer. Basically, a simple large and heavy heat sink seems to be better for the XHP50/70 LEDs at least.
- the wire gauge is unknown.
- stabilizing for 60 sec could be too long depending on the cooling solution. For lower power (30-60 Watts or so) this is not a huge deal, but if the thermal capacity is exceeded the light flux will go down rapidly due to rising temps inside the LED.
There is a table of maximum overcurrent of widely used Cree emitters. The XP-G4 (my test here) is given with 6 A, the XM-L2 new gen (my test here) is given with 10.25 A. The XHP70.3 HI has a max current of 4.65 A(12V, so for 6V it’s roughly doubled, at around 9.3 A or so), what is way beyond the theoretical maximum I achieved in my test.
So still relatively conservative, since they only tested to 85 °C Tsp. But still it could be possible that their cooling solution limits the maximum performance but I think they already considered that while designing the testing process. Keep in mind, that the Tj is somewhat higher due to thermal resistance of the board. So 85 °C Tsp are safe conditions for sufficient service life.
Unfortunately Cree does not give any light flux chart like I use for their testing…