OpenAI launches GPT-Rosalind, a reasoning model dedicated to life sciences and drug discovery

OpenAI launches GPT-Rosalind, a reasoning model for drug discovery and biological research tested by Amgen, Moderna, Thermo Fisher, and the Allen Institute.

OpenAI unveils GPT-Rosalind, a reasoning model designed specifically for biological research, drug discovery, and translational medicine. Named in honor of Rosalind Franklin, whose work helped reveal the structure of DNA, the model is optimized for complex scientific workflows covering chemistry, protein engineering, and genomics. The model addresses a structural problem in the pharmaceutical industry: the average ten-to-fifteen-year delay between the identification of a therapeutic target and the regulatory approval of a drug. GPT-Rosalind is designed to accelerate upstream stages, including scientific literature synthesis, hypothesis generation, experimental planning, and multi-omics data analysis. On the BixBench benchmark, dedicated to real-world bioinformatics, it demonstrates the best performance among available models. OpenAI is also releasing a Life Sciences plugin for Codex, available on GitHub, providing access to over fifty public databases and biological tools. The model is being deployed in restricted access for qualified companies in the United States, with a verification process covering research legitimacy, internal governance, and the prevention of misuse. Among the partners already involved are Amgen, Moderna, Thermo Fisher Scientific, and the Allen Institute.