I've been driving up to San Francisco this week from the south bay which has provided a nice test for the new .139 release. Long story short, if torque sleep is working, the impact is not perceptible - or at least it is not affected by the range mode setting.
A bit more detail - for those not in the bay area, going from Mountain View to San Francisco on 101 is a 40 mile mostly a flat drive. On my old S85, a drive to the city would probably be in the range of 300Wh/mile depending on speed. On Monday, when I drove up (on firmware .113, range mode off), I was getting roughly 330 Wh/mile. During the time that I drove there was little traffic, and I mostly used TACC at around 70-72 mph.
Last night I got the .139 update, and did the same drive to the city - this time with range mode on. Again, little traffic, TACC on at 72mph. I averaged about 328 Wh/mile. Well within the likely measurement error of my test, definitely not a 10% improvement.
During the ride I switched range mode off a couple of times over a 2 mile period to see if on the energy consumption chart I could see a difference - no change. Another poster on TMC suggested a perceptible difference in motor noise - I did not pick this up, though 101 has a very hard surface and is quite noisy with 21" wheels.
Overall, I'm not surprised. I would expect Tesla to confirm in the release notes the fact that torque sleep is in fact active. If this is the famed torque sleep release, then we're all in for some big disappointments.
One nice thing - I'm glad to see efficiency in the 325-330 Wh/m range. I don't do a lot of freeway driving and my average on my D has been 398 Wh/m over the 1200 miles I've had it (vs 348 for my S85 over 19k miles). Moreover, on 280 (same route as 101 but much more hilly), the efficiency didn't reduce that much. On my ride home last night, I got 345 Wh/m which is not bad. It seems that the D regens a lot more power than the S85 did.