OpenAI has rolled out GPT-Rosalind, a large language model fine-tuned for biology tasks. It’s named after Rosalind Franklin and trained specifically on the 50 most common biological workflows—everything from genome assembly to protein interaction mapping. The team also fed it data on how to query key public databases, so it can pull in high-quality information on genes, proteins and pathways. Yunyun Wang, OpenAI’s Life Sciences Product Lead, says GPT-Rosalind tackles two big headaches for researchers. First is data overload: decades of sequencing and biochemistry have produced petabytes of raw data. Second is jargon siloing: a plant geneticist often can’t easily wade through neurobiology papers when they spot a brain-expressed gene. This model bridges those gaps by translating across subfields and summarizing massive datasets. Beyond summarizing, GPT-Rosalind suggests likely biological pathways and ranks possible drug targets. It uses known regulatory mechanisms to link genotypes to phenotypes, and it can predict structural or functional protein features. The result should be faster hypothesis generation and better prioritization of experiments—especially in drug discovery and synthetic biology.