At present, the information dissemination model based on the structure modeling of online social network can no longer simulate the increasingly complex information dissemination process. Therefore, a heat transfer (HT) information dissemination model based on heat transfer idea is proposed in this paper. Firstly, based on the physical characteristics of heat transfer due to temperature difference between objects, the model introduces the temperature value attribute of nodes and defines the dissemination node and observer node. Secondly, the nodes in the online social network were described as heat receivers and dispersers. The diffusion heat and the receiving heat process of information dissemination were defined, and a two-stage greedy diffusion (TGD) algorithm based on HT model was proposed for information dissemination in online social network. Finally, the HT model is verified by experiments in real data sets. The experimental results show that the HT model proposed has high accuracy in the process of information dissemination of social networks of different scales and has great advantages in the prediction of dissemination trend and node variation trend.