The field of affective computing is rapidly advancing, with a growing focus on multimodal emotion understanding and the integration of large language models (LLMs) into various applications. Recent developments have highlighted the potential of LLMs for cross-lingual, audio-visual-textual emotion recognition, as well as their ability to generate human-like conversation and provide personalized therapeutic music retrieval. The use of LLMs in operations research has also shown promise, with applications in automatic modeling, auxiliary optimization, and direct solving. Furthermore, the development of high-quality datasets, such as the Affective Air Quality dataset and the MNV-17 dataset, has facilitated research in emotion recognition and nonverbal vocalization detection. Noteworthy papers in this area include the introduction of the EmoHeal system, which delivers personalized therapeutic music retrieval, and the survey on large language models and operations research, which outlines possible research avenues for advancing the role of LLMs in OR. The LLMs4All review also provides a comprehensive overview of the integration of LLMs into various academic disciplines, highlighting their potential impacts on research and practice.