OpenAI is offering GPT‑Rosalind to US labs—what is actually free?
Eligible university labs can apply for up to 30 no-fee seats for 12 months. Approval, usage limits and a US-only launch narrow the offer, and no study yet shows that the workbench improves research outcomes.
The short version
- What happened
- Eligible US university labs can apply for up to 30 Business Standard seats at $0 per user per month for as long as 12 months.
- Who it affects
- Eligible researchers may be able to test a costly category of tool across a lab without paying seat fees, while colleagues outside the United States remain excluded from the initial programme.
- What remains unproven
- Both sources are OpenAI materials. They provide product and eligibility terms but no controlled comparison, adoption denominator, error audit, independent replication or evidence of better scientific outcomes.
Keep in mind
Both sources are OpenAI materials. They provide product and eligibility terms but no controlled comparison, adoption denominator, error audit, independent replication or evidence of better scientific outcomes.
Multi-source analysis · 2 linked sources. Primary analysed source date: 15 October 2026.
Read the full report · 3 min The evidence, methods, implications and limitations
The offer removes a subscription price, not every constraint
OpenAI has opened applications for a Rosalind Academic Plan that offers eligible US university laboratories up to 30 Business Standard seats at $0 per user per month for up to 12 months. A principal investigator or lab lead applies, and access to GPT‑Rosalind—the company's research workbench—depends on a trusted-access review. The official help page says setup still requires a payment card, lab members must use their institution email domain, usage limits apply and API credits are not part of the programme.[1][2]
Why it matters for working researchers
For an eligible lab, 30 no-fee seats could lower the barrier to testing AI-assisted literature work, data analysis or experiment planning across a whole team. Research data is not used to train OpenAI's models by default under the stated business-data terms. But the initial US-only scope excludes international collaborators and many smaller institutions, while plan limits may matter for computationally intensive work. The offer funds access to a product, not research staff, wet-lab materials, independent validation or unlimited computing.[1][2]
The distinction that matters
What a lab must pass through before the no-fee trial begins
The advertised price applies only after eligibility and trusted-access checks.
Apply
A US university PI or lab lead submits the lab
Eligibility is reviewed
Approve
OpenAI completes trusted-access and account setup
Up to 30 institution-linked seats
Use
Researchers work within plan limits for up to 12 months
No seat fee; no API credits or proven research gain
This is an access announcement, not an effectiveness study
OpenAI's announcement describes tasks such as combining scientific literature, analysing data and planning experiments, accompanied by examples from academic researchers. Those accounts can show how people choose to use the workbench, but they provide no denominator, comparator or independently measured outcome. The company does not report how many labs tested it, how often outputs were wrong, whether researchers finished work faster, or whether resulting experiments reproduced.
Labs with an existing OpenAI Enterprise agreement are directed to their account team rather than the standard application route. Approval and usage rules also mean '$0 per user' should not be read as open, universal or unlimited access. A meaningful assessment would compare teams with and without the tool on prespecified tasks, record researcher time and correction burden, and test whether analyses, code and experimental decisions remain reproducible. Until then, the material development is narrower but real: selected US academic labs can trial a specialised frontier-model workspace without a seat subscription for one year.[1][2]