The field of natural language processing is moving towards more sophisticated and nuanced understanding of language, with a focus on long-context understanding and multimodal learning. Researchers are developing new benchmarks and evaluation methods to assess the capabilities of large language models, including their ability to process and comprehend extended contexts, perform deep reasoning, and understand implicit information. These efforts aim to improve the accuracy and controllability of language models, particularly in tasks such as summary generation, question answering, and discourse tracking. Notable papers in this area include: LC-Eval, which introduces a bilingual multi-task evaluation benchmark for long-context understanding, DiscoTrack, a multilingual LLM benchmark for discourse tracking, AcademicEval, a live long-context LLM benchmark for evaluating LLMs over long-context generation tasks, M3-SLU, a new multimodal large language model benchmark for evaluating multi-speaker, multi-turn spoken language understanding.