OpenAI offers GPT-Rosalind to academic biology labs
OpenAI is giving eligible university labs up to 30 sponsored seats for GPT-Rosalind, its specialized model for biological discovery workflows.
OpenAI has introduced the Rosalind Academic Plan, offering eligible academic laboratories sponsored access to GPT-Rosalind and the Rosalind Workbench. The program is aimed at life-science research involving medicinal chemistry, genomics, wet-lab assistance and related discovery tasks.
Each qualifying lab can receive up to 30 Business Standard seats at no cost for as long as 12 months. GPT-Rosalind access remains subject to trusted-access approval, while the plan also includes commercially available frontier OpenAI models within usage limits. API access is not included at launch, and the initial program is limited to academic labs at US universities.
OpenAI describes GPT-Rosalind as a specialized reasoning model rather than a general chatbot. Its stated use cases include designing and optimizing drug and protein candidates, studying disease signatures, exploring vaccines and immune responses, developing genetic and RNA therapies, and investigating antimicrobial resistance. The company says research data shared through the plan is not used to train its models by default.
The launch also illustrates how OpenAI is trying to put frontier models inside scientific institutions without treating access as completely unrestricted. Applicants must verify institutional legitimacy, the principal investigator’s identity and authority, the lab’s research activity and its designated safety contact. OpenAI says it will continue monitoring use and apply additional checks to high-sensitivity work.
Why it matters: the program lowers the entry cost for university groups that cannot buy enterprise access, while tying advanced biological reasoning to institutional review. Its significance will depend less on the number of free seats than on whether researchers can turn model-assisted literature review and hypothesis generation into experimentally validated results. The US-only scope and absence of API access also limit its immediate global reach.