This paper presents a comprehensive survey of Scientific Large Language Models (Sci-LLMs) in the biological and chemical domains. The authors highlight the growing importance of Sci-LLMs in scientific discovery and their potential to bridge the gap between natural language and specialized scientific languages. Sci-LLMs are designed to understand, interpret, and generate scientific languages, including textual, molecular, protein, genomic, and multimodal languages. The survey systematically reviews the technical advancements of Sci-LLMs, focusing on their architectures, capabilities, datasets, and evaluation methods. The authors analyze the challenges and future directions of Sci-LLMs, emphasizing the need for a unified framework to address the unique requirements of scientific languages. The survey also provides an overview of existing Sci-LLMs, including their applications in medical, biological, chemical, and comprehensive domains. The authors discuss the importance of scientific knowledge embedded in natural languages, such as textbooks, patents, and research papers, and explore the potential of multimodal LLMs to integrate various scientific languages. The survey emphasizes the need for further research to improve the performance of Sci-LLMs in scientific tasks, including text generation, prediction, and analysis. The authors also highlight the importance of open-source resources and collaborative efforts in advancing the field of Sci-LLMs. The survey concludes with a discussion of the future directions of Sci-LLMs, including the development of more efficient and accurate models for scientific tasks. The authors emphasize the importance of interdisciplinary research and the need for continued innovation in the field of Sci-LLMs.This paper presents a comprehensive survey of Scientific Large Language Models (Sci-LLMs) in the biological and chemical domains. The authors highlight the growing importance of Sci-LLMs in scientific discovery and their potential to bridge the gap between natural language and specialized scientific languages. Sci-LLMs are designed to understand, interpret, and generate scientific languages, including textual, molecular, protein, genomic, and multimodal languages. The survey systematically reviews the technical advancements of Sci-LLMs, focusing on their architectures, capabilities, datasets, and evaluation methods. The authors analyze the challenges and future directions of Sci-LLMs, emphasizing the need for a unified framework to address the unique requirements of scientific languages. The survey also provides an overview of existing Sci-LLMs, including their applications in medical, biological, chemical, and comprehensive domains. The authors discuss the importance of scientific knowledge embedded in natural languages, such as textbooks, patents, and research papers, and explore the potential of multimodal LLMs to integrate various scientific languages. The survey emphasizes the need for further research to improve the performance of Sci-LLMs in scientific tasks, including text generation, prediction, and analysis. The authors also highlight the importance of open-source resources and collaborative efforts in advancing the field of Sci-LLMs. The survey concludes with a discussion of the future directions of Sci-LLMs, including the development of more efficient and accurate models for scientific tasks. The authors emphasize the importance of interdisciplinary research and the need for continued innovation in the field of Sci-LLMs.