Just to bring a slightly different perspective, but I really like Fenix for how they measure and report the performance of their flashlights. In a nutshell, Fenix is the most honest of the bunch.
One of the disapointing trends in flashlights is "turbo" modes. Where the flashlight will step down after a period of time and continue running at reduced output. A large part of this is how the ANSI FL1 tests are performed. Turn on flashlight, wait two minutes. Measure output. That is the output you can claim. Now wait for the light to drop to 10% output. That is the runtime you can claim.
A sneaky manufacture could make a light turn on at 1000 lumens, wait two minutes, then drop to 101 lumens. Hold that 101 lumens for 30 hours. Now they can claim 1000 lumen runtime for 30 hours. This is the most extreme situation and anyone would agree that 1000 lumens for 30 hours is at the very least misleading.
Plenty of companies do this to a smaller extent. Most Eagletac flashlights will drop 20% after three minutes of power on. Now 20% is not too noticable and Eagletac has the decency to tell you what is going on. But why not just hold the higher output level? 20% is not going to make a huge difference thermally, maybe it'll shorten the LED's life by a month. 20% does give an edge in runtime numbers.
Nitecore uses a substancial turbo mode on their midrange products. They also don't mention it. It causes some head-scratching if you look at the specs. For example, consider the MT2C. It claims 360 lumens for 3 hours on high and 180 lumens for 4.25 hours on medium. You'd think the medium mode should run for at least 6 hours with that sort of performance on high. Instead the high drops 33% after three minutes.
Even Surefire has started doing this, with the latest updates to the Outdoorsman. They do not mention output drops 15% after two minutes.
So, Fenix. Fenix is outstandingly honest. Almost none of their products use turbo modes. A handful do. For example, the HL30. It has a turbo, and Fenix does not mention this. But that is okay! First, because the stepdown takes five minutes instead of two minutes. And more importantly because Fenix only claims 200 lumens for 1 hour 40 minutes. This is the runtime you would get if every five minutes you bumped the light back up to 100% output. No part of FL1 requires you go to these lengths and most companies would simply let the output drop and claim a three hour runtime instead.
Armytek deserves some kudos as well. Some of their lights step down (and they don't tell us how much) but the step down does not happen for 20 minutes. This is long enough to be useful and ten times longer than simple FL1-abuse would dictate. (It is linked to battery voltage instead of time and could be even longer.) Streamlight and Olight frequently provide runtime graphs in their manuals that clearly show any loose regulation or turbo stepdown.
Hopefully more companies will see the light and stop exagerating their performance. Icon was notorious for having extremely aggressive stepdown, and look where they ended up.
As an aside, it would be nice if companies (or FL2) provided runtimes for 90%, 50% and 10% output. This is enough information to have a good idea what sort of regulation is being used.
And of course much thanks to Selfbuilt, Subwoofer, Bigmac and all the other reviewers who do runtime plots.