The predictability of the sea surface temperature (SST) in seasonal forecast systems is crucial for accurate seasonal predictions. In this study, we evaluate the prediction of SST in the Global Seasonal forecast system version 5 (GloSea5) hindcast with particular interest over the western North Pacific (WNP) in which the SST can modify atmospheric convection and the East Asian weather. GloSea5 has a cold SST bias in the WNP that grows over at least 7 months. The bias originates from the surface heat flux in which the latent heat flux bias shows the biggest contribution. We identify the overestimated cloud in the first few days after initialization that causes insufficient shortwave radiation and negative bias of the surface net heat flux. Uncoupled ocean model experiments infer that the ocean model is unlikely the primary source of the SST bias.