Our marine-hull research programme asks which vehicle carries a survey payload furthest per kWh. For the glider-versus-propeller-AUV question we had been sweeping the glider's electronics load over 0.7 to 5 W, because we had no measurement.
A published Seaglider energy table (dive 227 of sg564, Iceland Sea, 1000 m, 11 hours) changes that: 5410 J in total, a 137 mW mean. Everything except the buoyancy engine, including controller, sensors and GPS/Iridium, came to 3777 J, about 0.095 W, below the range we swept.
Re-running the comparison with that load, the glider goes further than the AUV in 63 to 83% of the swept cases when the AUV spends 0.5 to 1 W on hotel load, and in all of them at 5 W. The break-even AUV hotel load has a median near 0.4 to 0.5 W. The decisive unsourced input is now the AUV's hotel load, not the glider's.
Caveats: it is a Seaglider with Slocum aerodynamics in the model, and one deliberately low-power, slow dive. Our earlier result, with a much larger 14 to 35 kJ dive energy from a Slocum battery test, was far less favourable to the glider, and we do not yet know how much of that difference is the vehicles. The log entry is in research/marine-hull/LOG.md (milestone 4d), and the script is experiments/run_seaglider_budget.py.