The UK's Nature website recently reported on the largest biological artificial intelligence (AI) model to date, Evo 2. This model is trained on data from 128000 genomes and has the ability to write entire chromosomes and small genomes from scratch, opening up a new path for designing completely new genomes. On the basis of the previous generation model Evo 1, scientists from the American Institute of Arc Research, Stanford University, and NVIDIA collaborated to develop Evo 2. Compared to Evo 1 trained on 80000 genomes of bacteria, archaea, and viruses, Evo 2 has significantly increased its training data volume, covering 128000 genomes. These genomes extensively involve humans, animals, plants, and eukaryotes, encompassing 9.3 trillion nucleotides. The team stated that given the enormous computing power required to parse these data and other features, Evo 2 is the largest biological AI model released to date. The Evo 1 and Evo 2 models, as important achievements in the emerging field of generative biology, will further enhance our understanding of the fundamental components of life. In the fields of healthcare and new drug development, the Evo 2 model can help understand gene variants associated with specific diseases. In the BRCA1 gene variant test related to breast cancer, Evo 2 has an accuracy rate of more than 90% in predicting benign mutations and potential pathogenic mutations. This will help save a lot of time and accurately design new molecules targeting these variants. In addition, in the field of agriculture, the Evo 2 model can provide new insights into plant biology, helping to develop more climate adaptive or nutrient rich crop varieties, and contributing to solving the global food shortage problem. In the field of materials science, the Evo 2 model can be used to design biofuels or break down proteins from petroleum and plastics. The team stated that Evo 2 has been opened to researchers around the world, who can easily use the model through the website or download the source code, training data, and parameters of the model for free to explore the mysteries of biology together. (New Society)