OpenAI is offering university labs free access to GPT-Rosalind, its life sciences model, for up to a year. Under the new Rosalind Academic Plan, eligible labs can receive up to 30 Business Standard seats at no cost for up to 12 months, subject to OpenAI's trusted-access approval. For research groups that cannot fit frontier AI tools into their grant budgets, that removes the main barrier to trying one.
The plan bundles the model with Rosalind Workbench, which connects it to tools for medicinal chemistry, genomics, wet-lab assistance and other scientific work. OpenAI pitches GPT-Rosalind at deeper reasoning on complex biological questions, from designing candidate drugs, therapeutic proteins, peptides and enzymes to tuning how molecules interact, and says academic labs are already using it for work ranging from protein design to neurodegeneration. Approved labs also get access to OpenAI's other frontier models included in the plan.
The fine print matters. The plan is currently limited to labs at US universities. Principal investigators at universities without an OpenAI Enterprise agreement apply through OpenAI's trusted-access process, which reviews their professional background, appointment and authority over the lab; those at universities with an agreement apply through their institution. Seats go to named lab members, the lead manages permissions and is responsible for approved use, and monitoring continues after activation. API access is not included, which limits labs that want to build GPT-Rosalind into their own pipelines.
The offer is part of a wider push. In the same weeks, OpenAI showed thousands of GPT-Rosalind agents analysing a mouse brain CRISPR atlas and bought the drug development platform Orchestra Bio. Anthropic, for its part, has opened 8 to 12 week guest research projects for life scientists with compute at no cost. The two labs are competing for the same academic users and the same showcase results.
For researchers, the practical question is what happens next. Free seats are a familiar way to win habits, and after 12 months a lab that has built its methods around the tool will face a price or a negotiation. OpenAI's announcement does not say what the plan costs afterwards, how lab data is handled beyond its general terms, or how the trusted-access review treats sensitive biological work. Labs should read the data and usage terms closely, keep their methods portable where they can, and treat the year as a trial rather than a commitment.