Toyota's official numbers for the 3.0-liter turbocharged inline-six cylinder sourced from BMW are 335 hp and 365 lb-ft, measured at the crank. Using the hoary conventional wisdom of driveline and hydraulic losses causing 15 percent of power to evaporate before reaching the wheels, we'd expect the Supra's power at the contact patch to be around 285 hp and 310 lb-ft. Even updating that parasitic loss figure to 10 percent for a modern drivetrain, we're still at around 300 hp and 328 lb-ft.
Using MT's tallies and parasitic losses of 10 percent, the Supra in question was putting out more like 364 hp and 426 lb-ft at the crank. At 15 percent loss, the numbers suggest 390 hp and 445 lb-ft. Those numbers don't line up with the higher-powered incarnation of the B58 engine in the Z4 M40i, either, officially classified at 380 hp and 369 lb-ft.
The C/D torque figure can be rule an anomaly, the mag measuring torque in fifth gear at a 1.32:1 ratio instead of the 1.00:1 ratio in sixth gear because they couldn't hit redline in sixth. The mag avowed "there's a chance the peak torque output may be slightly inflated." On the other hand, Mustang dynos like the kind used by MT have earned the nickname "heartbreaker" from a habit of returning disappointing power totals.
Between car, region, weather, tester, and dyno variations, it's useless to directly compare numbers worked up by C/D and MT. The point is that two data points arrived at independently found noteworthy power surpluses compared to Supra brochure numbers.
Neither story tries to assign a reason for the difference. C/D, and the Internet, say BMW is known for understating outputs. When C/D put the current M5 Competition on the same Dynojet used to test the Supra, the readout showed 617 hp and 606 lb-ft at the wheels for a sedan officially rated at 617 hp and 553 lb-ft. MT said its press loaner was like an "early build preproduction car," perhaps with a "partially hand-built engine" and without the latest calibration.
Check out both articles for more details on the testing, and videos. Not that it matters for the moment, though; the real test will be to see whether customer cars get extra oomph, too.