The recent advancements in the research area of multimodal information retrieval and extraction are significantly pushing the boundaries of what is possible with large language models (LLMs) and generative AI. The field is witnessing a shift towards more sophisticated and versatile frameworks that can handle complex, unordered, and multi-modal data inputs. Innovations in input reranking for LLMs, generative AI-powered Monte Carlo methods for complex queries, and the repurposing of large multimodal models for retrieval tasks are leading the charge. These developments are not only enhancing the accuracy and efficiency of retrieval systems but also enabling them to generalize better across various tasks and modalities. Additionally, there is a growing emphasis on multilingual and multitask datasets, which are crucial for advancing multimodal information extraction. These datasets are facilitating the development of models that can integrate and leverage multimodal information more effectively, even in scenarios with missing or incomplete data. Overall, the field is moving towards more robust, adaptable, and high-performing systems that can handle a wide array of retrieval and extraction tasks with unprecedented precision and reliability.