I did a little testing using my ToolKitRC servo tester. It reads out the current the servo is using in real time.
I clamped the servo to my bench and suspended 675g from a 20mm servo arm. I recorded the idle amps with no load, the amps to lift the weight and the idle current with the weight attached.
MKS 747: idle 28mA, idle w/ load 200mA, moving 675mA
Futaba BLS173SV: idle 26mA, idle w/ load 26mA or 300mA*, moving 960mA
Savox SV1261MG: idle 10mA, idle w/ load 310mA or 10mA*, moving 1,700mA
Promodler DS100DV; Idle 10mA, idle w/ load 531mA or 10mA*, moving 950mA
The Futaba would go to idle current with the load attached. The Savox and Promodler will too, but I had to play with the output signal. The MKS would not go to idle current with a load. This may be due to how the control algorithm is written between different brands.
It leads me to think the MKS will use more energy over time.
The Savox & MKS are rated at the same torque, but the Savox drew more current to move the same load.
I like the way the Futaba went to idle amps and quit buzzing with the load attached. I'm not sure they are strong enough for flaps on a 14lbs jet.