OpenAI has unveiled GPT-Rosalind, its first domain-specific AI model purpose-built for life sciences research, marking a decisive move by the company into the high-stakes world of drug discovery and biological reasoning. The frontier model, announced on April 17, is now available as a research preview in ChatGPT, Codex, and the API for qualified enterprise customers through OpenAI's trusted access program.
Named after Rosalind Franklin -- the British chemist whose X-ray crystallography work was instrumental in revealing the double-helix structure of DNA -- GPT-Rosalind is designed to reason across molecules, proteins, genes, pathways, and disease-relevant biology. It represents what OpenAI calls the "first release in our Life Sciences model series," signaling that more specialized scientific models are on the way.
What GPT-Rosalind Can Do
Unlike general-purpose large language models, GPT-Rosalind is optimized for complex, multi-step scientific workflows. The model supports evidence synthesis, hypothesis generation, experimental planning, literature review, sequence-to-function interpretation, and data analysis -- tasks that typically consume weeks or months of a researcher's time.
Alongside the model, OpenAI is releasing a freely accessible Life Sciences research plugin for Codex that connects to more than 50 scientific tools and data sources spanning human genetics, functional genomics, protein structure, biochemistry, clinical evidence, and study discovery. Critically, while GPT-Rosalind itself is restricted to trusted-access customers, the Codex plugin is available to anyone using OpenAI's mainline models, significantly broadening the reach of the scientific tooling.
Benchmark Performance
On BixBench, a benchmark designed for bioinformatics and data analysis, GPT-Rosalind achieved a pass rate of 0.751, surpassing competitors including Gemini 3.1 Pro and Grok 4.2. On LABBench2, the model outperformed GPT-5.4 on 6 of 11 tasks, with its strongest gains in CloningQA. Perhaps most impressively, in real-world RNA prediction tasks conducted in partnership with Dyno Therapeutics, the model scored at the 95th percentile on prediction tasks and the 84th percentile on sequence generation using novel, unpublished sequences.
Industry Partners Line Up
OpenAI is not launching GPT-Rosalind in a vacuum. The company has assembled an impressive roster of collaborators including Amgen, Moderna, the Allen Institute, Thermo Fisher Scientific, Los Alamos National Laboratory, Dyno Therapeutics, and Novo Nordisk. The Novo Nordisk partnership, announced just days earlier on April 14, will deploy OpenAI's AI technologies across the Danish pharmaceutical giant's operations from drug discovery to commercial, with full integration targeted by year-end.
"Our unique collaboration with OpenAI enables us to apply their most advanced capabilities and tools in new and innovative ways, with the potential to accelerate how we deliver medicines to patients," said Sean Bruich, who leads AI strategy at Amgen.
Moderna CEO Stephane Bancel offered a broader perspective on the model's significance: "GPT-Rosalind represents an important step in helping scientific teams use advanced AI to reason across complex biological evidence."
Access Restrictions and Governance
GPT-Rosalind is not a model anyone can simply sign up for. The trusted-access deployment structure limits initial availability to qualified enterprise customers in the United States. Participating organizations must demonstrate that they are conducting legitimate scientific research with public benefit, maintain compliance and misuse-prevention controls, and show strong security governance. OpenAI has built checks around eligibility, access management, and organizational governance -- a notably cautious approach for a company that has faced criticism over the pace of its deployments.
Analysis: A Strategic Pivot with Real Stakes
The launch of GPT-Rosalind represents more than a product announcement -- it is a strategic statement about where OpenAI sees its future. By moving into domain-specific models for life sciences, the company is competing directly with specialized biotech AI firms like Recursion Pharmaceuticals, Insilico Medicine, and Isomorphic Labs, the Alphabet-backed venture led by Demis Hassabis.
The 50-plus tool integrations and the Codex plugin suggest OpenAI is building not just a model but an ecosystem -- one that could become a default platform for computational biology research. If GPT-Rosalind delivers on its benchmarks in real-world settings, the implications for drug development timelines could be substantial. The average new drug currently takes 10 to 15 years and over 2 billion dollars to bring to market; shaving even a fraction off that timeline would represent enormous value.
As Sam Altman put it: "AI is reshaping industries, and in life sciences, it can help people live better, longer lives."
The cautious rollout through the trusted access program also signals a maturing approach to deployment. By restricting access to vetted organizations with governance controls, OpenAI appears to be learning from past controversies about releasing powerful capabilities too broadly. Whether this restraint holds as competitive pressure mounts remains to be seen.
What Comes Next
OpenAI has framed GPT-Rosalind as just the beginning of a life sciences model series. With partners already applying the technology across discovery pipelines and a free Codex plugin democratizing access to scientific tooling, the foundation is laid for a broader push into specialized AI for research. The question is no longer whether AI will transform drug discovery -- but which platform will own the workflow.