Abstract: The existing emergency rescue systems are plagued with issues such as inefficient analysis of disaster information and a lack of humanization in the human-machine collaboration process. Large language models（LLM）with robust natural language understanding and multimodal comprehension abilities are expected to tackle the above difficult problems. By surveying the latest research progress in large models and sorting out and analyzing the mission requirements for LLM in emergency rescue and control field, the current technological maturity of LLM for application in the emergency rescue field is evaluated. The potential LLM application construction plans and a comprehensive construction plan for five task scenarios, including disaster information analysis and understanding, etc. are discussed. Finally,the future impact of LLMtechnology on the development of emergency rescue systems is prospected.