OpenAI Offers Free ChatGPT Seats to Academic Labs, With Conditions

OpenAI introduced the Rosalind Academic Plan on October 15, 2026, for eligible academic labs at U.S. universities. Labs that qualify get up to 30 complimentary ChatGPT Business Standard seats for up to 12 months. The price during that period: $0 per user per month.

The seats bundle access to every commercially available frontier OpenAI model within the plan’s usage limits. A Standard seat includes ChatGPT and Codex, but those tools keep their existing usage limits. The offer does not remove those caps.

Notably, the plan does not include API credits. A payment card is required during setup, even though there is no charge for the 12-month period.

GPT-Rosalind Is Separate

The headline model here is GPT-Rosalind, OpenAI’s dedicated life sciences model. It comes with the Rosalind Workbench, which has two modes: Explore for general scientific questions and Research for complex biological problems. GPT-Rosalind can also be used inside Codex.

But access is not automatic. Plan approval does not grant GPT-Rosalind through the API. Instead, labs need separate approval through OpenAI’s trusted access process. The principal investigator or lab lead must then enable it individually for each member in admin settings.

Data and Workspace Rules

OpenAI says it does not use research data shared through the plan to train its models by default. However, there is no Zero Data Retention under this offer. Standard data retention applies in a managed ChatGPT Business workspace.

The verified PI or lab lead is the sole workspace owner. They cannot appoint additional owners or admins. The workspace is capped at 30 seats.

Who Can Apply

Eligibility is narrow: academic labs at U.S. universities only. PIs or lab leads without an OpenAI Enterprise agreement apply through the trusted-access process. Universities that already have an Enterprise agreement request via their Enterprise access process.

The application requires signing in with a ChatGPT account tied to an institutional .edu email. Applicants must submit:

• Institution, department, and lab name
• Official website
• PI profile, role, and research focus
• Intended use
• Names and institutional emails of people needing access
• A trained biosafety contact at the institution (must be an institution-designated official)

Review uses Persona for identity verification. A university email or existing subscription does not guarantee approval.

Once approved, PIs can activate free access for up to 30 named members. All must be on the same institutional email domain as the PI. Visiting researchers or collaborators from other institutions cannot be added. Unaffiliated researchers and commercial organizations are not eligible.

Research Areas Covered

The plan lists seven research areas:

• Drug and protein design
• Disease detection and monitoring
• Vaccines and immune responses
• Genetic and RNA therapies
• Agriculture and animal health
• Ecology and conservation
• Antimicrobial resistance

What OpenAI Showed as an Example

The announcement included an access example. A prompt requested sequence-level designs for a second-generation Nipah antiviral campaign. GPT-Rosalind produced a candidate comparison matrix, a modeling plan, and study plans.

A general model, called Mainline Astra, withheld sequence-level designs. It said such details would provide actionable assistance involving high-consequence pathogens.

OpenAI notes that the assessment addresses textual fulfillment, not scientific correctness or feasibility. In other words, the model produced a structured answer, but that does not mean the science is sound.

Early Users Named

OpenAI named three early academic users:

• UCSF Zhao Lab (quantitative drug development)
• Bourdenx Lab (led by Mathieu Bourdenx)
• Stanford/Arc Institute’s Brian Hie (discovery and design tasks at larger scale)

Why OpenAI Framed It This Way

OpenAI positioned the plan around academic research’s role in earlier advances. The announcement cited the Human Genome Project, bacterial immunity research that led to CRISPR-Cas9, and decades of university mRNA research behind COVID-19 vaccines.

The message: academic labs did foundational work before, and this plan aims to give them tools for the next round.

What This Means in Practice

For eligible labs, the offer is straightforward: free access to frontier models for a year, no training on your data by default, and a clear path to GPT-Rosalind with extra approval.

But the constraints are real. The 30-seat cap is per single workspace. No extra admins. No outside collaborators. No API credits. No Zero Data Retention. And the biosafety contact requirement adds an institutional layer that some labs may not have ready.

The separate approval for GPT-Rosalind means the plan is not a one-click unlock. PIs must request trusted access, then enable the model per member. That is a deliberate gate, likely tied to the sensitive nature of life sciences work.

The Nipah example is telling. GPT-Rosalind gave sequence-level designs; a general model refused. That contrast shows why OpenAI built a dedicated model and why it restricts access. The plan gives academic labs a foot in the door, but the high-consequence pathogen work stays behind an extra check.

For labs that qualify, once past the card setup hurdle and the biosafety paperwork, the value is clear: a year of free frontier model access. For those that do not fit the strict eligibility, the plan is not an option.

OpenAI did not say whether the plan will expand beyond U.S. universities or to non-academic research institutions. For now, the Rosalind Academic Plan is a U.S.-only, academic-only, PI-gated offer.