OpenAI's GPT-Rosalind: A New Era for Life Sciences AI

OpenAI said it is introducing a new series of artificial intelligence models aimed at speeding up work in life sciences, including drug discovery and translational medicine. The company announced the lineup on Wednesday, April 16, and said the tools are designed to help researchers handle large and complex datasets used across genomics, protein analysis, and biochemistry.

At the center of the release is GPT-Rosalind , a model OpenAI said is built for foundational reasoning in biochemistry and genomics. The model is named after chemist Rosalind Franklin. OpenAI positioned the system as a way to synthesize evidence, generate hypotheses, and support analysis, while emphasizing that it is intended to augment scientists rather than replace them.

OpenAI linked the effort to the long timelines and high failure rates that characterize drug development. The company cited a 10–15 year timeframe for drug development and approval, and said that only one in ten drugs that enter clinical trials ultimately receives approval. It also pointed to the impact on millions of people globally living with rare diseases, framing the models as part of an attempt to make research and translation to therapies more efficient.

The company said it will begin a controlled rollout through a research preview, using what it called a trusted access program . OpenAI said the preview will be available to select enterprise customers starting Thursday, naming Amgen , Moderna , the Allen Institute , and Thermo Fisher Scientific among the initial participants. OpenAI also said the models will be offered with enterprise-grade security controls.

The announcement comes alongside ongoing concerns about how AI trained on biological data could be misused, including for creating dangerous pathogens. OpenAI noted that its controlled rollout is intended to maximize beneficial use while reducing risks. Separately, an international group of over 100 scientists has called for tighter controls on sensitive biological data used in AI systems.

OpenAI also highlighted limits in the current state of the field, saying that few AI-discovered or AI-designed drugs have progressed beyond early clinical trials. That caveat underscores that the company is presenting the models as research tools rather than a guaranteed shortcut to approved medicines, even as it argues they can help scientists navigate the scale and complexity of modern biological data.