Same emitter, same battery, different run times. Who has the best drivers?

Is there a way to get more detailed information on what drivers are used in our favorite brands and how efficient they may be?
Here's a little list of lights with all the same XM-L emitters and tested with a "2600 mah" 18650 battery. Maybe the Olight and 4Sevens tested with 2x123?

Sunwayman T20C: High: 438 Lumens (2hrs) Medium: 70 Lumens (15hrs) Low: 10 Lumens (50hrs)
Olight M20-X: High: 500 Lumens (1.5hrs) Medium: 100 Lumens (5hrs) Low: 10 Lumens (85hrs)
4Sevens Maelstrom X7: High: 480 Lumens (1.3hrs) Medium: 270 Lumens (2.8hrs) Low: 30 Lumens (21hrs) Moonlight: 0.3 Lumens (144hrs)
Klarus XT11: High: 470 Lumens (2.3hrs) Medium: 150 Lumens (7.3hrs) Low: 10 Lumens (300hrs)

All these numbers are provided by the manufacturers and ANSI rated.

So what makes the Klarus low mode outperform the 4Sevens moonlight mode? What makes the Olight less efficient in medium mode but have a more efficient low mode than the Sunwayman?

Runtime profiles. Look for runtime charts and you'll see that some lights maintain 'good' regulation, remaining near peak output (and draining the battery faster) until the end. Others slouch to give longer runtime at almost-as-bright levels.

It had become standard practice here on CPF to list runtimes to 50%, certainly the great Flashlight Review site started by CPF member Quickbeam did it that way many years ago. Most others that followed here do it that way too, such as Selfbuilt. I think it is a MUCH more useful statistic than runtime to 10% of starting output, which I believe is the ANSI standard. THis means that a light can tail off very significantly and still "keep going" as far a the runtime rating is concerned. Real life usefulness would be very different toward the end of those times for some of those lights. Don't get too caught up in those numbers without actually seeing the runtime graph, as AnAppleSnail mentioned.

Thrifty

+2 on seeing the runtime graph - that is the only way to know what illumination a light actually puts out over time.

My prediction is that over time as the new ANSI standards become widely accepted some manufacturers will tweak their drivers so that they put out a lot of lumens initially, then once past the measuring point for ANSI Max lumens (120 seconds) they will then drop off quite quickly to give a very long tail (just above the 10% used for ANSI Run time). This will make the figures on the box (and in advertising) look much better than a more evenly regulated light that may have lower official ANSI figures but put out a lot more light on a runtime graph. I see this as a bad thing.

On the other hand it may encourage manufacturers to supply better quality primary lithium batteries with their lights, as this will boost the the max lumens figure on some lights. I see this as a good thing.

The 4Sevens has a rotating head UI, which requires some energy to constantly control the sensors. Infinite variable selector ring lights have the same problem. There are also many other electronic design characteristics, like input voltage range, PWM or constant current, which influence the runtime.

I'm not reallytrusting the Ansi specs form some manufacturers. The Klarus MiX6 says Low is 3lm and Med is 26 lm but actual values are more like 0.8lm and 18lm.

Check out the reviews for the best info.