The field of molecular generation and verification is rapidly advancing, with a focus on developing innovative methods for designing and evaluating molecules with desired properties. Recent developments have centered around improving the efficiency and effectiveness of molecular language models, as well as enhancing the verification and validation of generated molecules. Notably, zero-knowledge proof mechanisms are being introduced to ensure the structural privacy of molecular outputs, while advances in flow matching and diffusion-based models are enabling the generation of high-quality molecules with improved validity and diversity. Furthermore, research is exploring the application of large language models to protein function prediction and amyloidogenicity prediction, demonstrating promising results. Overall, these advancements are paving the way for significant breakthroughs in molecular generation and verification, with potential applications in drug discovery and other fields. Noteworthy papers include: ToxiEval-ZKP, which introduces a structure-private verification framework for molecular toxicity repair tasks, and ProtTeX-CC, which enhances protein large language models through two-stage instruction compression. FlowMol3 is also notable for its advancements in flow matching for 3D de novo small-molecule generation, and NovoMolGen for its systematic investigation of molecular language model pretraining. Additionally, Cross-Modality Controlled Molecule Generation with Diffusion Language Model is noteworthy for its ability to extend pre-trained diffusion models to support cross-modality constraints.