Accurate multi-physics modelling is necessary to simulate and predict the long-term behaviour of subsurface porous rocks. Despite decades of modelling subsurface multi-physics processes in porous rocks, there are still considerable uncertainties and challenges remaining partly because of the way the constitutive equations describing such processes are derived (thermodynamically or phenomenologically) and treated (continuum or discrete) regardless of the way they are solved (e.g. finite-element or finite-volume methods). We review here continuum multi-physics models covering aspects of poromechanics, chemo-poromechanics, thermo-poromechanics, and thermo-chemo-poromechanics. We focus on models that are derived based on thermodynamics to signify the importance of such a basis and discuss the limitations of the phenomenological models and how thermodynamics-based modelling can overcome such limitations. The review highlights that the experimental determination of thermodynamics response coefficients (coupling or constitutive coefficients) and field applicability of the developed thermodynamics models are significant research gaps to be addressed. Verification and validation of the constitutive models, preferably through physical experiments, is yet to be comprehensively realized which is further discussed in this review. The review also shows the versatility of the multi-physics models to address issues from shale gas production to CO2 sequestration and energy storage and highlights the need for inclusion of thermodynamically consistent damage mechanics, coupling of chemical and mechanical damage and two-phase fluid flow in multi-physics models.