In this paper, we report a correction to the model potential of the Ga acceptor in germanium, evidenced by high-magnetic-field photoconductivity measurements. We found that under high magnetic fields the chemical shift of the binding energy of Ga acceptors vanishes, contrary to the results given by the generally accepted theory. To fit our data, we found that the central-cell correction should contain a repulsive part (i.e., it must be bipolar), in contrast to the purely attractive screened point-charge potential widely used in the literature.