The viral gene sequence is regarded as a kind of special text, which constitutes a core innovation in the application of large language models to virus research. This idea breaks free from the shackles of the traditional analysis framework and opens up a new way of computing to understand the evolution of the virus. Using this method, researchers can quickly and accurately locate the key mutation sites of the virus. This helps to detect emerging variant strains in time and assess their potential impact on transmissibility and pathogenicity. Especially importantly, this technology can also reconstruct the evolutionary path of the virus and realize real-time tracking of the mutation process. This provides a powerful tool for a systematic understanding of the adaptive evolution of the virus. These profound insights based on the big language model can be used as a theoretical basis for the development of targeted vaccines and drugs. Although the relevant computing technology is in the development stage, it has shown great potential to promote the transformation of virological research to a predictable direction.