95 x 6 = 570 W
Assuming you run all channels at full
4 X 250 = 1000W
People tell me watts are watts so you already decrease your heat load by almost 50%
Due to the physics of light delivery (LED's being more directional and focused) you can approach a photon equivalency in the tank using less watts vs MH's.
Using quality diodes and their higher effeciency to begin w/ helps.
Something you would lose w/ cheaper less efficient diodes.
Balancing that a bit is the more diffuse nature of 4 mh's so in some areas, and until they catch up i.e deeper the distribution w/ LED's is less..
As to the arguement of using the sensible heat from the MH's (lot of emissions in the IR area) to compensate for resistive heaters in the tank.. well depends on room temperature differential vs tank temp for a big part of it.
Complicated ..
Power yes, effective photons, not so less,
If you dim a bunch of channels yes again
Think that's pretty much a wash..
Each radion will be responseable for roughly a 16x18 area and each mh a 24x36 area.
May slightly favor the LED
Depends on the corals and a whole bunch of other parameters.
If you as mentioned above, dimmed the LEd's likelihood of slower growth increases.
PURE OPINION based on physics of the lights not any definitive statement of success.
Don't expect 6 radions dimmed by 25% to push the same photon count as 4 full blast 250W MH's..regardless of diode efficiency and "normal" lensing. See even a single 3W led can push a ton of "par" if lensed to 1 degree-ish but in a spot the size of a nickle..

just physics..
Heat balance is really a case by case thing but one thing not arguable (in my mind) is most LED heat is up and away while a good chunk of the MH is down..
Also don't forget that to a certain degree on can make up for lack of instantaneous quantity by increasing the photoperiod..
I probably made more questions than answers here though..