To further clarify, the XM-L2 LED's should produce more output when driven by the same amount of power as an XM-L. This is their advantage - next generation, more lumens/per amp at the same voltage and same temperature, same package size. So, if the cost is the same and the bin, tint, and beam pattern&color is all the same, and they actually do perform better as stated, they would replace XM-L. There are a lot of "if's" there though 
and CREE doesn't have a stellar record with filling out the product family offerings, or filling them out quickly.
But if direct drive, say an XM-L will pull 3A, but an XP-G will only pull 1A. If my understanding is correct. Which would be valid to the original question. Basically asking if XM-L2 would handle more amperage or be more efficient with the same amperage.
Don’t beat me up to bad, I’m still a newb 
Direct drive is very silly way of driving a LED. Don’t do it.
I was just trying to simplify the equation by removing the driver.
Could you clarify why it is a bad idea?
Because, with LEDs, current varies massively even with small voltage difference. And it goes against manufacturer’s (of LED) recommendations.
Is that necessarily bad?
Yes, it’s not as efficient as driving an LED at 1A or 2A but we do get the peak brightness of an XM-L at 3-4A.
Plus, LED’s just don’t go poof at even 300%+ of their recommend current, specified by the manufacturer.
The HD2010, one of the most highly regarded budget lights, is in fact directly driven at High mode.
What’s wrong with driving at whatever current you want using a constant-current driver? Direct drive has no advantages over constant-current drivers at all.
Nothing at all, except linear driver using AMC 7135’s burn the excess power as heat.
Other than that, using a constant current drivers does enable you to control the brightness.
Linear drivers aren’t great either. Especially AMC7135 based ones - you can only regulate current in 350mA steps using them, so no option for moonlight and low modes without resorting to PWM.
Yup, I hope we can produce BLF style drivers, preferably a high-efficiency buck driver with BLF’s programmable codes. :party:
Synchronized buck/boost drivers would be best, for those who want to run 1 cell. Buck/*boost* because draining modern 2900/3100/3400mAh cells requires boosting due to them having not-so-high voltage.
Modern cells can’t stay over 3.3-3.4 volts under load?
Anyways, I wonder if we can arrange something with Fasttech about our drivers. /wistful thinking
Nope (at over 2A)
Not totally correct. P=UxI. Different LEDs have different Vf
If someone decides to organise group buy sign me in for 2 or 3 … or 4 
Mouser shipped mine out, Qty 3, XM-L2 T6's. We will see. I got 2 cheapo Sipik SK73 zoomies, 1 with a U2 1A, other with a stock T6. I can do ceiling bounce measurements on these 2, then swap the XML T6 for the new one, interesting comparison then...
I received the 7 xml2 this morning and quickly put them into my dry. I am loving the tint of these leds
33696933 - Wow! That was fast! Dunno about the DRY - single or multi LED? Was the form factor the same for mounting? Height the same? What's the tint comparable to warmer or cooler than a T6 3C? Can you notice brightness diff or can you do a ceiling bounce test w/meter?
Sorry - lots of Q's... You are the first one known to get, see, and install these things...
the dry is multi (3). it looks similar to a 3c to me
Noticeably brighter or the same?