OpenAI launches sponsored GPT-Rosalind plan for U.S. academic labs

OpenAI has launched the Rosalind Academic Plan, opening applications for approved academic labs at U.S. universities to receive sponsored access to ChatGPT Business and, after a separate review, GPT-Rosalind.

Approved labs may receive up to 30 ChatGPT Business Standard seats at $0 per user per month for up to 12 months. The offer excludes API access and API credits. Access to ChatGPT and Codex, frontier models and GPT-Rosalind remains subject to applicable usage limits.

A principal investigator or designated lab lead must apply. Labs at universities with an OpenAI Enterprise agreement must use their institution’s Enterprise access process instead. Neither an institutional email address nor an existing OpenAI subscription guarantees approval; applicants must meet the eligibility rules and pass a trusted-access review. OpenAI’s program guidance says a payment card is required during setup despite the complimentary seat price during the approved period.

The approved PI or lab lead becomes the sole owner of a managed ChatGPT Business workspace, controlling membership and model permissions. Joining the workspace does not unlock GPT-Rosalind by itself; the owner must enable the model for eligible members. Members must use verified institutional email addresses on the same domain as the approved lead, excluding collaborators whose addresses belong to another institution. The owner must remove access when a member leaves, changes affiliation or no longer needs it.

Before activation, OpenAI says it verifies the institution and the lead’s identity, reviews the lead’s authority, the lab’s affiliation, its research and intended use, and requires an institutional safety contact. After activation, the company says it may seek clarification about research, narrow permissions or revoke access if requirements are not met.

The company says research data shared through the plan is not used to train its models by default. Standard Business-workspace retention still applies, and Zero Data Retention is not available under the offer.

In its GPT-Rosalind-5.5 system card, OpenAI classifies the model as High capability, but below Critical, in its Biological and Chemical preparedness domain. The company describes safeguards combining trusted-access controls, model boundaries against harmful biological and cyber assistance, monitoring and enforcement, and additional contractual security requirements.

OpenAI says the plan could support discoveries and that advanced AI can accelerate early research workflows. Yet the launch materials provide no measured outcomes showing that the Academic Plan has accelerated discovery, improved research quality or produced scientific or clinical breakthroughs. OpenAI has not disclosed how many labs it will admit, the program’s total budget or an application deadline.