OpenAI's Rosalind Academic Plan Gives US Labs 30 Free Seats for 12 Months, Without API Access

OpenAI opened applications on Oct. 15 for the Rosalind Academic Plan, which gives eligible US university labs up to 30 free ChatGPT Business Standard seats for up to 12 months, according to AI Weekly and Crypto Briefing. The Terminal could not fetch OpenAI's own announcement, which returned an access error, so the terms below are as those two outlets report them.

The plan is the first general route into GPT-Rosalind for university research, AI Weekly reported. Until now the model was mainly for large companies and select government partners.

What a lab gets and what it does not

A principal investigator or lab lead applies for the team and manages approved members, AI Weekly reported. The seats come with ChatGPT and Codex, which support literature review, research data work and analysis code. Research data entered through the plan is not used to train OpenAI's models by default, per both outlets.

API access is not included, AI Weekly reported, so a lab cannot script GPT-Rosalind into a pipeline through this program.

Term	As reported
Seats	Up to 30 Business Standard, free
Duration	Up to 12 months
Who can apply	Eligible US university labs
API access	Not included
Model access	Separate trusted-access review

The vetting is the real gate

Approval requires institutional verification, a background review of lab leadership and a designated institutional safety contact, according to AI Weekly. GPT-Rosalind itself needs a separate trusted-access review. Crypto Briefing says access is not automatic and that OpenAI described the focus as legitimate scientific research, without giving further criteria.

OpenAI has not published, in anything The Terminal could read, how many labs it expects to admit or how long review takes.

What GPT-Rosalind is

Crypto Briefing describes GPT-Rosalind as OpenAI's life sciences model, built to reason across biological entities, synthesize evidence from the literature, plan experiments and connect to scientific databases. It launched as a research preview on April 16, 2026, limited to vetted enterprise and research organizations, with Amgen and Moderna named as early collaborators.

The outlet relays gains over baselines of 53.7 percent on Genebench, 18.0 percent on Medchem Bench and 19.6 percent on Labworkbench. Those are vendor-supplied figures and neither outlet says who ran them or against which baselines, so they cannot be compared with anything else here.

Crypto Briefing also reports that a model listed as gpt-rosalind-discovery appeared at $5 per million input tokens as of Oct. 10, and says it does not tie that listing to the API.

How this compares with other lab science claims

The plan arrives in the same month as other lab-science claims the site has covered. Anthropic said Claude cut quantum circuit gates 17x on one task, and OpenAI's own math output was counted at 722 papers, under 1 percent of them in statistics and optimization.

Neither outlet says when OpenAI will start approving labs, so the 12-month clock has no start date yet.