"OpenAI appears to be testing a new subscription and experience for science use cases, but it's unclear if it'll be available to everyone regardless of their background," reported BleepingComputer.
ChatGPT for Science subscription — what the leak shows
The leak, first noticed on X and visible in references on OpenAI's web build, points to a new offering called "ChatGPT for Science." It appears as a distinct subscription or model category alongside OpenAI's existing lines: ChatGPT personal, Teams, and business/enterprise. The page copy spotted in the test suggests a science-focused experience, but does not resolve a core question: whether the service will be open to individual researchers or restricted to institutional users.
How ChatGPT for Science would differ from existing subscriptions
The source material identifies clear differences among OpenAI’s current tiers: ChatGPT personal is designed to work for everybody; Teams requires a company domain and at least three users; ChatGPT business is restricted to legal entities. The leak implies ChatGPT for Science will likely follow a similar pattern of constraints, with the possibility that only verified institutes or universities could gain access. The defining characteristic advertised for the new tier is a stronger grounding in scientific discoveries and research compared with a regular subscription.
GPT-Rosalind and the enterprise-scale science model
The report places ChatGPT for Science in context with OpenAI’s recent work on specialized models. It notes GPT-Rosalind, built on the foundation of OpenAI’s GPT-5.5 architecture, as an example of a purpose-built model for life sciences. The source is careful to say GPT-Rosalind is not merely a reskinned ChatGPT with a science prompt; instead, it is a highly specialized model designed for enterprise-scale life sciences research and locked behind what OpenAI calls a "trusted-access deployment structure."
Trusted-access deployment structure and security posture
According to the leak, GPT-Rosalind is strictly available to eligible organizations — examples named include major pharmaceutical companies like Novo Nordisk and verified research institutions — that are conducting legitimate, public-benefit scientific research. The model requires enterprise-grade security and strong safety governance, standards the source says mirror or exceed the strict requirements of ChatGPT Enterprise. The report suggests OpenAI may be planning to bring some of these enterprise capabilities to a broader class of institutions through a ChatGPT for Science subscription rather than keeping them restricted to a few select partners.
What this means for technologists, verified research institutions, and pharmaceutical companies
- Technologists and security teams: will watch for the deployment and verification mechanisms; the source highlights enterprise-grade security and strong safety governance as prerequisites for access to specialized models like GPT-Rosalind.
- Verified research institutions and universities: are the likeliest audience for a ChatGPT for Science tier, since the leak repeatedly references verification and institutional eligibility rather than individual sign-ups.
- Pharmaceutical companies and large life-sciences organizations: already named as eligible for GPT-Rosalind (with Novo Nordisk given as an example), these organizations could remain within the trusted-access circle unless OpenAI chooses to broaden availability through the new subscription.
Testing status, timeline, and the outstanding question of access
The leak makes clear that ChatGPT for Science is being actively tested on the web build and that an announcement is likely weeks away. Beyond that near-term timetable, the principal unanswered fact in the material is whether OpenAI will open the science tier to all users or constrain it to verified institutional customers — the report repeatedly signals the latter as likely but stops short of confirmation.
OpenAI’s recent moves — a specialized model locked behind trusted access, paired with a leaked web test for a science subscription — frame a straightforward trade-off: specialized capabilities and heightened security governance on the one hand, and restricted access on the other. Whether ChatGPT for Science becomes a broadly available tool for individual researchers or remains a gated resource for verified organizations will determine who benefits from these science-specific AI advances.