We assessed the performance of state-of-the-art coupled models in reproducing the equatorial undercurrent (EUC) in the Indian Ocean based on the outputs of the Coupled Model Intercomparison Project Phase 6 (CMIP6) models and compared with the Phase 5 (CMIP5) models. Our results showed that the CMIP6 models reproduced the boreal winter–spring Indian Ocean EUC more realistically than the CMIP5 models, although both generations of models underestimated the strength of the Indian Ocean EUC compared with the observations. This underestimation of the Indian Ocean EUC can be attributed to the excessively strong and westward-extended cold tongue in the equatorial Pacific. In the CMIP models, a stronger winter-mean cold tongue favors a stronger zonal sea surface temperature gradient, which forces a strong easterly wind bias over the equatorial western Pacific. This, in turn, contributes to an acceleration of the Walker circulation. This enhanced Walker circulation over the Indo-Pacific Ocean directly causes a lower level westerly wind bias over the equatorial Indian Ocean and drives a shallow west–deep east thermocline tilt bias, ultimately leading to an excessively weak EUC in the Indian Ocean via wind-induced thermocline processes. Compared with the CMIP5 models, the overall improvement in the strength of the winter–spring Indian Ocean EUC in the CMIP6 models can be traced back to the improvement in the degree of the strong and westward-extended cold tongue bias. Our results suggest that efforts should be made to reduce the bias in the mean-state equatorial Pacific sea surface temperature to further improve the simulation and projection of the atmospheric and oceanic circulations in the Indian Ocean.