OpenAI releases GPT‑Rosalind

Biology is the science of how cells, genes, and proteins work, and drug discovery starts by finding where that machinery breaks. OpenAI said on April 16 it released GPT‑Rosalind, a model built for that research process rather than for general chat. (openai.com) OpenAI described GPT‑Rosalind as a “frontier reasoning model” for biology, drug discovery, and translational medicine, the stage where lab findings are turned into treatments. The company said the model is optimized for chemistry, protein engineering, genomics, and scientific workflows that use outside tools and databases. (openai.com) A protein is a folded chain of amino acids that acts like a tiny machine inside the body, and genomics is the study of DNA instructions that tell cells what to build. OpenAI said GPT‑Rosalind is meant to help researchers reason across those systems in multi-step tasks, including protein understanding, genomics analysis, and biochemistry. (help.openai.com) Drug discovery is slow: OpenAI said a new drug in the United States typically takes about 10 to 15 years to move from target discovery to regulatory approval. The company argued that faster work at the earliest stages can improve target selection, biological hypotheses, and experiment design before a molecule ever reaches a clinical trial. (openai.com) OpenAI is not offering GPT‑Rosalind as a broad public chatbot. It said the model is in research preview and is being deployed only to eligible institutions through a “trusted-access” process tied to safeguards against misuse. (openai.com) The release fits a wider OpenAI push into science tools. In December 2025, the company launched OpenAI for Science, saying it wanted to build AI systems that help researchers explore ideas, test hypotheses faster, and shorten discovery timelines across disciplines. (openai.com) OpenAI has also been building the measurement and safety pieces around that strategy. In 2025 it introduced FrontierScience, a benchmark for expert-level questions in physics, chemistry, and biology, and it separately published work on preparing for future AI capabilities in biology and the biosecurity risks they could create. (openai.com, openai.com) The company’s own materials frame the commercial opening in concrete terms: the PDF released alongside the announcement says drug development costs roughly double every nine years, while clinical trials take nearly 7.5 years on average. The same paper points to companies including Exscientia, LabGenius, and Recursion as examples of AI-assisted drug-discovery efforts already reshaping parts of the pipeline. (openai.com) For now, OpenAI is pitching GPT‑Rosalind as a specialist lab assistant, not a universal model. The next test is whether institutions that get access can show faster or better biology work outside OpenAI’s own announcement materials. (openai.com, openai.com)