The field of natural language processing is witnessing significant advancements in multilingual language models, with a growing focus on improving cross-lingual transfer, cultural alignment, and sociolinguistic diversity. Researchers are exploring innovative approaches to bridge the performance gap between high-resource and low-resource languages, including fine-tuning on synthetic code-switched text, using word association learning, and developing culturally grounded evaluation frameworks. These efforts aim to enhance the fairness and robustness of large language models in multilingual contexts, enabling more effective communication and understanding across linguistic and cultural boundaries. Noteworthy papers include: When Does Language Transfer Help, which investigates the effectiveness of sequential fine-tuning for cross-lingual euphemism detection. ALIGN, which introduces a cost-efficient approach to modeling and aligning culture in large language models using word association learning. Long Chain-of-Thought Reasoning Across Languages, which presents a systematic study of long chain-of-thought generation across multiple languages.