The field of natural language processing is witnessing significant advancements in the development of multilingual large language models (LLMs). Recent research has focused on improving the performance of LLMs on low-resource languages, with a emphasis on tokenization, language identification, and translation quality estimation. The use of multilingual encoders, adaptive layer optimization, and cross-prompt encoders has shown promising results in enhancing the capabilities of LLMs for low-resource languages. Furthermore, the application of LLMs in constructed language creation, immigration discourse analysis, and code-switched child-directed speech has demonstrated their potential in diverse areas. Noteworthy papers in this regard include ConlangCrafter, which introduces a multi-hop pipeline for end-to-end conlang creation, and TopXGen, which presents an LLM-based approach for generating high-quality and topic-diverse parallel data for low-resource machine translation. Overall, the field is moving towards more efficient, scalable, and equitable multilingual LLMs, with a focus on improving performance on low-resource languages and exploring new applications.