The field of structure-based drug design and protein modeling is moving towards the development of more sophisticated generative models that can accurately incorporate spatial and structural constraints. These models aim to improve the design of drug ligands and proteins by leveraging advances in Bayesian flow networks, diffusion-based generators, and neural fields. Notably, researchers are focusing on addressing key challenges such as incorporating boundary condition constraints, integrating hierarchical structural conditions, and ensuring spatial modeling fidelity. The use of novel frameworks and architectures is enabling the generation of more accurate and diverse molecular structures, including peptides and macrocyclic peptides. Overall, the field is advancing towards more automated and data-conditioned protein modeling, with potential applications in drug discovery and development. Noteworthy papers include: SculptDrug, which proposes a spatial condition-aware Bayesian flow model for structure-based drug design. Seek and You Shall Fold, which introduces a framework for non-differentiable guidance of protein generative models. Full Atom Peptide Design via Riemannian Euclidean Bayesian Flow Networks, which presents a Bayesian flow network for full atom peptide design. Unified all-atom molecule generation with neural fields, which introduces a framework for generating target-conditioned, all-atom molecules across atomic systems.