Foundation models, such as CNNs and ViTs, have powered the development of image modeling. However, general guidance to model architecture design is still missing. The design of many modern model architectures, such as residual structures, multiplicative gating signal, and feed-forward networks, can be interpreted in terms of the heat conduction equation. This finding inspired us to model images by the heat conduction equation, where the essential idea is to conceptualize image features as temperatures and model their information interaction as the diffusion of thermal energy. We can take advantage of the rich knowledge in the heat conduction equation to guide us in designing new and more interpretable models. As an example, we propose Heat Conduction Layer and Refine Approximation Layer inspired by solving the heat conduction equation using Finite Difference Method and Fourier series, respectively. This paper does not aim to present a state-of-the-art model; instead, it seeks to integrate the overall architectural design of the model into the heat conduction theory framework. Nevertheless, our Heat Conduction Network (HcNet) still shows competitive performance. Code available at https://github.com/ZheminZhang1/HcNet.