The field of multimodal understanding and reasoning is rapidly evolving, with a focus on developing more sophisticated models that can effectively integrate and process multiple forms of input, such as text, images, and other modalities. One of the key challenges in this area is the ability to reason about complex, nuanced concepts, such as humor, metaphor, and misinformation. Recent work has explored the use of large language models and neuro-symbolic frameworks to address these challenges, with promising results. Notably, researchers are making progress in developing models that can detect and understand humorous multimodal metaphors, as well as those that can effectively integrate material and formal reasoning. Additionally, there is a growing interest in developing explainable and trustworthy models, particularly in the context of misinformation detection. Noteworthy papers include: Multimodal Reference Visual Grounding, which introduces a novel method for visual grounding that leverages reference images to improve object detection. PEIRCE, which proposes a neuro-symbolic framework for unifying material and formal inference. EXCLAIM, which presents a retrieval-based framework for detecting Out-of-Context misinformation. Do Reasoning Models Show Better Verbalized Calibration, which investigates the calibration properties of large reasoning models.