OpenAI makes GPT-6 Rosalind available for biology research
OpenAI has made a new version of Rosalind available to approved organisations working in biology and drug discovery. Access is limited because of risks associated with some applications of biological research.
Summary
OpenAI has made GPT-6 Rosalind Research available to a limited group of approved organisations conducting life sciences research. The model is intended to support work in areas including biology, drug discovery and translational medicine, which aims to move research findings towards health applications.
The new version was released on 30 October 2026. Access requires specific approval for both the organisation and its users; a ChatGPT subscription alone is not sufficient.
In practice
According to OpenAI, the Rosalind family can help teams analyse biological data, review scientific literature, work with protein and genomic information, and plan experiments. GPT-6 Rosalind Research is available to approved organisations through Work and Codex and, with separate approval, through the company’s API.
Approval for Rosalind Discovery, a tier intended for more routine scientific tasks, does not automatically grant access to Rosalind Research. OpenAI also says the new version is not available in ChatGPT’s general Chat tab.
Context
GPT-Rosalind-6 uses the same model weights as GPT-6 Astra. OpenAI says it did not add training data or carry out additional training for this release. The difference lies in how the system is made available and the safeguards applied to biological research.
The company says Rosalind refuses fewer requests on certain dual-use biological topics: knowledge that can support legitimate research but may also have harmful applications. Before granting access, OpenAI says it reviews applicants’ intended uses, security controls and governance practices.
In its own assessment, OpenAI rates the system’s biological and chemical capability as high, below the company’s critical threshold. That assessment does not, by itself, demonstrate the discovery of medicines or the effectiveness of treatments.
Why it matters
- The model gives some scientific teams access to a tool designed to support complex research tasks.
- Restricted access is intended to balance scientific usefulness with the risk of biological knowledge being misused.
- The claimed benefits will need to be assessed through research outcomes, as well as the provider’s own tests.