OpenAI announced the launch of GPT-Rosalind, a purpose-built reasoning model designed to support scientific research across biology, drug discovery, and translational medicine. The new model is optimized for complex life sciences workflows, combining advanced reasoning capabilities with improved integration across scientific tools, databases, and experimental processes.
The introduction of GPT-Rosalind reflects the growing need to address inefficiencies in life sciences research, where the process of developing a new drug can take 10 to 15 years from initial discovery to regulatory approval. Early-stage improvements in hypothesis generation, target selection, and experimental design can significantly impact downstream success rates, but these workflows remain fragmented and time-intensive.
GPT-Rosalind is designed to help researchers navigate these challenges by enabling faster evidence synthesis, hypothesis generation, experimental planning, and data analysis. The model supports multi-step reasoning across domains such as chemistry, genomics, and protein engineering, while also helping scientists uncover connections that may otherwise be overlooked.
The model is now available as a research preview through ChatGPT, Codex, and the API for qualified users under a trusted access program. OpenAI is also introducing a Life Sciences research plugin for Codex, which connects researchers to more than 50 scientific tools and data sources, enabling more efficient execution of common workflows such as literature review, sequence analysis, and dataset discovery.
OpenAI is collaborating with leading organizations including Amgen, Moderna, Thermo Fisher Scientific, NVIDIA, the Allen Institute, and UCSF School of Pharmacy to apply GPT-Rosalind across real-world research workflows. These partnerships aim to accelerate discovery and improve the quality of scientific outputs in areas ranging from molecular biology to clinical research.
In benchmark evaluations, GPT-Rosalind demonstrated improved performance across core scientific domains, including chemical reasoning, protein structure analysis, genomics, and experimental design. On the BixBench benchmark for bioinformatics tasks, the model achieved leading performance among comparable systems, while also outperforming prior models on multiple tasks in the LABBench2 evaluation suite.
The model also showed strong results in industry testing. In collaboration with Dyno Therapeutics, GPT-Rosalind ranked above the 95th percentile of human experts on RNA prediction tasks and around the 84th percentile on sequence generation tasks when evaluated in controlled settings.
To ensure responsible deployment, GPT-Rosalind is being rolled out under a controlled access framework. Organizations must meet criteria related to beneficial use, governance, and security, and access is restricted to approved users operating in secure research environments. OpenAI emphasized that the system includes enterprise-grade safeguards to mitigate risks associated with biological misuse.
Looking ahead, OpenAI plans to expand the GPT-Rosalind model series, improving its ability to handle long-horizon and tool-intensive scientific workflows. The company is also working with research institutions, including Los Alamos National Laboratory, to explore applications such as AI-guided protein and catalyst design.
The model is named after Rosalind Franklin, whose work was instrumental in uncovering the structure of DNA, reflecting OpenAI’s broader ambition to build AI systems that accelerate scientific discovery and contribute to advancements in human health.
KEY QUOTE:
“The life sciences field demands precision at every step. The questions are highly complex, the data are highly unique, and the stakes are incredibly high. 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.”
Sean Bruich, Senior Vice President of Artificial Intelligence and Data, Amgen
