According to foreign media reports, OpenAI has added the "gpt-rosalind-discovery" model to the life sciences model category in its API pricing table. The model was not announced at a separate launch event, nor accompanied by a blog post or system card; it was simply quietly listed in the pricing table, with the information first shared by an AI tracking bot on the X platform.
OpenAI adds gpt-rosalind-discovery model to expand life sciences applications
According to the pricing table, GPT-Rosalind Discovery's input price is US$5 per 1 million tokens (approx. HK$39), cached input US$0.50 (approx. HK$4), output US$25 (approx. HK$195), identical to the existing gpt-rosalind-research. Judging from the model name, it is likely used for discovering new drug candidate molecules, though OpenAI has not disclosed specific use cases or availability scope, and the context window size has not been confirmed.
GPT-Rosalind is a biology, drug discovery, and translational medicine reasoning model released by OpenAI in April, with Amgen, Moderna, the Allen Institute, and Thermo Fisher Scientific as early partners. On September 11, OpenAI ended the model's research preview phase and began making it available to qualified institutions worldwide through the Trusted Access Program. The pricing table shows that the research version began billing on October 5, with usage currently limited to approved internal research.
The latest version, GPT-Rosalind-5.5, is based on GPT-5.5. The system card shows that this model has been specially trained not to refuse difficult biology questions, while using access restrictions as the primary safety measure. The OpenAI Help Center describes it as most useful in early research and hypothesis-building stages such as target biology, literature reviews, and omics data interpretation. Omics refers to the field of large-scale analysis of biological molecules such as genomes and proteomes.