For the H50 they specified the lumens on the H50 for NiMH - they did not specify the battery assumed for the H501. With the higher efficiency of the 14500 you could drive the LED a little harder get a higher lumen count and still get the same runtime? They might even have a more efficient driver on board?
But actually, it does look somewhat odd...
I looked at the numbers for the H50 and H501 and they seem quite believable to me. I think no one mentioned it earlier but 66 lumens for 2.33 hour for the H50 doesn't seem to be really that efficient. ZL probably now simply has a better behaving circuit.
If I did my math correctly, to achieve 66 lumen output the H50 would need around 0.25A on the LED.That would be 3.12V Vf and around 0.65A taken from a NiMH. Multiplied by 2.33 hours runtime it would give around 1.5Wh needed for a light with a perfectly efficient circuit. A 2700 mAh cell holds approximately 3Wh. That would mean that the H50 circuit consumes 3Wh when the perfect one would need 1.5Wh. In other words, the overall H50 efficiency in the high mode would be only around 0.5. Quite an interesting place for improvement here, I think.
With the similar calculations one would see that for the H501 such an efficiency figure is around 70%-75% (96lm, 2.3h, 0.36A on the LED, 3.17Vf, 0.95A on the battery, 2.3Wh needed for a perfect circuit).
These are only really approximate calculations and we would not know for sure before someone performs a full runtime test. But I believe they show that the advertised 96lm for 2.3h seems to be achievable. And these figures are definitely much closer to the truth than those provided by some other manufactures who claim to be able to drive the LED with twice the amount of the energy than they have in an AA NiMH cell.