June 4, 2026
OpenAI announced new capabilities for GPT-Rosalind on June 3, 2026, its model purpose-built for life sciences research at enterprise scale. The updated GPT-Rosalind combines GPT-5.5’s agentic coding and tool use with stronger performance in drug-discovery domains such as medicinal chemistry and genomics, while using 31% fewer tokens than GPT-5.5. OpenAI also launched Rosalind Biodefense, an initiative for defensive applications of AI in the life sciences.
What is GPT-Rosalind?
GPT-Rosalind is an OpenAI model purpose-built for life sciences research at enterprise scale. It is named after Rosalind Franklin, the researcher whose work was essential to understanding the molecular structures of DNA, RNA, and viruses. The model targets research tasks across biology, medicinal chemistry, genomics, and laboratory workflows, rather than general-purpose chat.
The June 3, 2026 announcement is an update that adds new capabilities to the existing GPT-Rosalind model. It carries over GPT-5.5’s agentic coding and tool-use abilities and pairs them with stronger domain knowledge in core drug-discovery areas.
What is new in the GPT-Rosalind update?
The updated GPT-Rosalind improves model intelligence in core drug-discovery domains, including medicinal chemistry and genomics, and extends performance across broader life sciences analysis, design, and experimental workflows. In OpenAI’s evaluations, the model shows broad gains on research tasks set by biology experts, complex medicinal chemistry queries, quantitative biology, and wet lab troubleshooting.
A practical addition is the model’s ability to link perturbations to experimental outcomes in real wet lab protocols. This can be used to troubleshoot failed experiments and to optimise existing protocols, which connects the model’s reasoning to concrete laboratory results rather than to text alone.
GPT-Rosalind performance and token efficiency
The headline efficiency figure is that GPT-Rosalind uses 31% fewer tokens than GPT-5.5 while improving accuracy on the evaluated research tasks. Fewer tokens for the same or better result lowers the cost and latency of long research workflows, which matters for the multi-step analysis and tool use that life sciences research typically requires.
OpenAI reports the gains across several categories at once: expert biology tasks, medicinal chemistry, quantitative biology, and wet lab troubleshooting. The combination of GPT-5.5’s agentic coding with stronger domain intelligence is what OpenAI positions as the main advance in this release.
How does GPT-Rosalind compare to GPT-5.5?
GPT-Rosalind is built on top of GPT-5.5’s agentic coding and tool-use foundation, so it inherits those general capabilities. The difference is specialisation: GPT-Rosalind adds stronger intelligence in drug-discovery domains and improves accuracy on life sciences research tasks while using 31% fewer tokens than GPT-5.5.
Where GPT-5.5 is a general-purpose model, GPT-Rosalind is scoped to life sciences and delivered through a trusted-access deployment structure rather than as a broadly available product. For organisations doing biology, chemistry, and genomics research, GPT-Rosalind is the targeted option; for general work, GPT-5.5 remains the relevant model.
GPT-Rosalind availability and Rosalind Biodefense
GPT-Rosalind is available in research preview to eligible organisations globally through OpenAI’s trusted-access deployment structure. Access is therefore limited rather than open, in line with the sensitivity of life sciences capabilities.
Alongside the model update, OpenAI launched Rosalind Biodefense, a new initiative to enable high-impact defensive applications of AI in the life sciences using GPT-Rosalind. The initiative is framed around strengthening societal resilience and pandemic preparedness through defensive biodefense work.
For full details on the GPT-Rosalind update and the Rosalind Biodefense initiative, see the official announcement on the OpenAI website.
