eng-0124
window recorded
citable URL: https://halobench.com/records/eng-0124/ — this address never moves; the anchor /records/#eng-0124 keeps resolving
130.7 Whmean 136.58 W − 10.1 W idle floor (box-idle-all-empty) = 126.48 W attributable
window 2026-08-16T09:11:47Z → 2026-08-16T10:09:11Z · provenance recorded
method wall-meter · aihydra energy sensor via Home Assistant (counter-difference on sensor.hardware_ai_hydra_energy, raw ~10s-resolution history samples, linearly interpolated to the run-meta window) · no contention
per task 7.2588 Wh · 0.2199 p at 30.3 p/kWh
covers run-0329
aihydra laguna-s21-fullbench tau2 26-task run window (09:11:47Z-10:09:11Z, 3444s), matching run-0329 exactly (excludes the preceding 73s server-load window, 0.64 Wh / 31.7 W, per the deepseek-v4-flash convention -- see run-0329's note). Counter interpolated from raw ~10s-resolution HA history samples pulled fresh this session: start 21.811893394767356 kWh, end 21.94255229834369 kWh. Recomputed independently from HA raw history (not copied from the delegate's dead-drop report), reproducing its headline arithmetic: 130.66 Wh / 136.6 W mean. wh_per_task is Wh-per-CORRECT-ANSWER: wh_total / tasks_passed (18 of 26, run-0329.metrics) = 7.2588 Wh, 0.2199 pence at the standing 30.3 p/kWh tariff. WHOLE-SESSION energy -- includes simulator (user_llm) wait time end to end, not an agent-only or decode-only figure, directly comparable to deepseek-v4-flash's eng-0107 and nemotron3-super's eng-0116 on the same denominator.
Wh, never joules — read as a cumulative-counter difference across the window; delta over the idle baseline is the only figure that means anything
Cited by — computed at build time, never stored
model pages laguna-s-21