OpenAI has unveiled a significant upgrade to its life sciences AI model, GPT-Rosalind, combining the intelligence of GPT-5.5 with enhanced capabilities across drug discovery, genomics, and medicinal chemistry. The update represents the company’s most ambitious effort yet to bring frontier AI directly into the hands of the world’s leading scientific researchers.
The improved model demonstrates impressive performance on specialized scientific benchmarks. On MedChemBench, GPT-Rosalind outperforms GPT-5.5 while using fewer computational resources. On GeneBench it achieves higher accuracy on complex biology tasks, and on LabWorkBench it scores significantly higher on real wet lab experimental outcomes.
Named in honor of Rosalind Franklin — the pioneering chemist whose X-ray crystallography work was essential to uncovering the structure of DNA — the model is built to work across biological scales, from molecules to genes to entire living systems.
OpenAI is also expanding the model’s practical reach through new Life Sciences plugins for its Codex platform, enabling bioinformatics workflows and evidence-backed biological interpretation.
The company has partnered with major organizations including Novo Nordisk, Amgen, Moderna, the Allen Institute, and Thermo Fisher Scientific to deploy the technology across real-world research pipelines, and the model is now available to eligible organizations worldwide.
