Some GPUs have fully separate 12V rails, others just short the two together. In the first case, load on the connector would depend on which board loads are being powered by each connector (ex.: RAM + fans + RGB + 4 Vcore phases on one, 5 Vcore on the other, 1 Vcore from slot power) and whichever cable feeds the baseline load will get 10-20W more light-load load than the other. When the two cables are parallelled on the board, the load distribution between the two will vary depending on slight differences in wiring and board resistance and unless something has gone horribly wrong, the imbalance shouldn't be more than a few %.
One is more than double the other, the difference is almost certainly caused by the load split between the two.