If you ever get around to doing this comparison between the different colored LED's, note that different colors have different voltage requirements. Red, orange, yellow have the lowest minimum input voltage (just over 2V, I believe) while the other colors require just over 3V. As a given coin cell gets run down, its voltage will gradually drop and you'll reach a point where the cell can't provide the minimum voltage. So, that cell might light up a red Photon but not a blue, even though you're using exactly the same, run down cell in both Photons. That's why some colors use two cells while other colors still come with just use one cell.

Based on the voltage requirements, I think you're pretty much guranteed that the red, orange, yellow LED's will outlast the others if they are all powered by a single 2032.

But! Not to discourage you from doing some of your own hands-on experimenting! Maybe my educated guess isn't quite correct in real world use. Have at it, Alex!

There are good products out there like this. My general rule of thumb is that if it isn't a 'Freeplay' device, it's not worth my money.

This is the most recent product I've seen. I've seen devices with Freeplay technology in the form of flashlights, radios, and cellphone chargers...they've been around a while.

I have to agree with JohnN and Hacksaw regarding the Gerber Infinity. I really like this light. The lithium batteries certainly cut down the weight and extend the runtime, but even with a standard (or a depleted) AA this is a nice really light to have on your person. The unit is also legitimately waterproof. The finish on the aluminum case is pretty easily worn when the light is in your pocket with other items. Strictly a cosmetic issue. Also, I was concerned about the exposed bulb (albeit recessed), however, that has not proven to be an issue at all even when carried in a pocket with small hard objects.

Alex, I really look forward to the results of your runtime tests in a chilled environment.

By God, I completely forgot the power save mode...Very good catch there! Now that would make things interesting!

Factors I want to work in, now.

The power save mode, thanks again for reminding me
Cold environment to simulate less than ideal conditions, thinking of keeping it in the fridge or even a snow-drift if I get the lights going in the winter. May run a room temperature test on a whim later on.
All using 2032 cell, various colors. If things die quickly on some colors, may substitute two 2016s in future tests.
Simply lighting up is not enough, must provide functional illumination.
Straight run through, if it goes out it's out of the running.
Further testing of colors to determine if the longer running ones are even usable in the actual field.
Constant on, no strobe modes.

Could you try at least one white 2x2016 for comparison, since that is what the majority of people will use?

Are you aware of the runtime tests done on the Candlepower forums? They would be interested in your results too.