OpenAI has introduced the Rosalind Academic Plan for academic labs seeking access to its specialized life-sciences model and associated workbench. The publisher's announcement sets out who can apply, what the sponsored access includes and which responsibilities remain with the lab.
Check institutional eligibility first
The plan is currently available to academic labs at U.S. universities. Principal investigators and lab leads at institutions without an OpenAI Enterprise agreement can apply on behalf of their labs through the trusted-access process.
Labs at universities with an existing agreement must use their university's Enterprise access process. That distinction gives a lab a concrete first step: establish which institutional route applies before creating a separate application.
Understand the sponsored access
Eligible labs can receive up to 30 Business Standard seats for up to 12 months. The listed cost is zero per user per month during that sponsored period, with access to commercially available frontier models subject to the plan's usage limits.
Access to GPT-Rosalind requires trusted-access approval. The announcement describes it as a dedicated life-sciences model used alongside Rosalind Workbench, with researchers retaining scientific expertise and judgment in their workflows.
API access is not included. A lab planning an automated integration should account for that exclusion rather than assuming the sponsored seats supply a programmable API allowance.
Prepare for verification and ongoing oversight
OpenAI describes checks of the institution, the lab lead's identity and authority, research activities and intended use. The process also establishes responsibilities for the lead and a designated institutional safety contact.
The announcement says research data shared through the plan is not used to train models by default. Approved lab leads manage named members and their permissions, while monitoring and enforcement continue after activation. These are conditions of the announced access program, rather than evidence that model output independently validates a scientific result.