OpenAI Tests ChatGPT for Science: New Subscription for Research Institutes and Universities Discovered in Web Code

A leak in the web client of ChatGPT reveals that OpenAI is testing a new subscription called ChatGPT for Science, aimed at verified research institutes and universities. References to the plan have been found directly in the web build code, without any official communication from the company.

Based on the indications that have emerged, ChatGPT for Science would offer specialized guidance on scientific discoveries and literature compared to a standard subscription. Access would likely be restricted to verified entities, similar to the Teams plan, which requires a minimum of three users with a company domain, but extended to academic institutions and research centers with institutional verification.

The direct context is GPT-Rosalind, a model for life sciences enterprise recently announced by OpenAI, built on GPT-5.5. Rosalind is accessible only through a distribution framework reserved for eligible organizations: large pharmaceutical companies like Novo Nordisk and verified research institutes conducting activities for the public good. This access excludes the vast majority of universities and academic labs.

ChatGPT for Science seems to address this gap: OpenAI may want to bring some of Rosalind's capabilities to all institutions through the new subscription, rather than restricting them to select partners. It is still unclear whether the underlying model will be GPT-5.5 or a variant, nor what specific features would be included in the more open plan.

Current ChatGPT plans include four configurations: Personal, Teams, Business, and Enterprise, with the latter reserved for legal entities. ChatGPT for Science would fit as a fifth vertical option, the first specifically designed for the academic and research world.

The launch date is unknown. The plan is in active testing on the web, and according to indications, an announcement could arrive within a few weeks.