OpenAI announced GPT-Rosalind, a frontier reasoning model purpose-built for biology, drug discovery, and translational medicine. Named after Rosalind Franklin, it's optimized for scientific workflows and combines improved tool use with deeper understanding across chemistry, protein engineering, and genomics.
It's available as a research preview in ChatGPT, Codex, and the API for qualified customers through OpenAI's trusted access program. Launch partners include Amgen, Moderna, the Allen Institute, and Thermo Fisher Scientific.
A Codex Plugin Anyone Can Use
Alongside the model, OpenAI shipped a Life Sciences research plugin for Codex, available for free on GitHub. The plugin provides a modular set of skills for common research workflows—human genetics, functional genomics, protein structure, biochemistry, clinical evidence, public study discovery—and exposes more than 50 public multi-omics databases, literature sources, and biology tools as an orchestration layer.
Eligible Enterprise users can pair the plugin with GPT-Rosalind for deeper biological reasoning, but all users can use it with mainline OpenAI models. For developers building scientific tooling, this is essentially a reference implementation of a domain-specific skills pack on the Codex plugin platform.
Performance Numbers OpenAI Highlighted
GPT-Rosalind achieved leading published-score performance on BixBench (real-world bioinformatics and data analysis) and beat GPT-5.4 on 6 out of 11 tasks on LABBench2—with the largest gain on CloningQA, which requires end-to-end DNA and enzyme reagent design for molecular cloning protocols.
In a partnership with Dyno Therapeutics on RNA sequence-to-function tasks using unpublished, uncontaminated sequences, best-of-ten model submissions evaluated directly in the Codex app ranked above the 95th percentile of human experts on the prediction task and around the 84th percentile on sequence generation, compared against 57 historical scores from AI-bio researchers.
Trusted Access and Safety Controls
The model is launching through a trusted-access deployment for qualified Enterprise customers in the U.S. first, with controls around eligibility, access management, and organizational governance. OpenAI evaluates access on three principles: beneficial use, strong governance and safety oversight, and controlled access with enterprise-grade security. During the research preview, use of the model does not consume existing credits or tokens, subject to abuse guardrails. Pricing details will follow as the program expands.
What It Means for Web Developers
Two takeaways for developers outside life sciences. First, the architecture is the news: a domain-specialized model plus a public skills/plugin pack that orchestrates ~50 external tools is a template OpenAI is likely to repeat for other verticals. Second, the Codex plugin ecosystem just got a flagship example—if you're building tools that need to reason over many APIs and databases, the Life Sciences plugin on GitHub is a working pattern to study, regardless of whether you ever touch a protein structure.