OpenAI has launched its first domain-specific AI model for biology and drug discovery, named GPT-Rosalind. The model, designed to accelerate early-stage research workflows, outperforms its predecessor on specialized benchmarks but is tightly restricted to U.S. enterprise users due to biosecurity concerns. Major pharmaceutical firms including Amgen and Moderna are among the initial customers.
OpenAI has unveiled GPT-Rosalind, a specialized reasoning model targeting biology, drug discovery, and translational medicine. It represents the first release in the company’s new Life Sciences model series, entering a competitive field alongside other specialized labs.
The model is named after British chemist Rosalind Franklin, whose work was crucial to understanding DNA’s structure. OpenAI announced the model on Thursday, aiming to compress the 10 to 15 years typically required for U.S. drug approval.
On the BixBench benchmark for real-world bioinformatics tasks, GPT-Rosalind achieved a top published score of 0.751. It also outperformed GPT-5.4 on six out of eleven tasks in the LABBench2 evaluation.
OpenAI’s life sciences research lead, Joy Jiao, provided measured expectations for the model’s capabilities. “We do think there’s a real opportunity to help researchers move faster through some of the most complex and time-intensive parts of the scientific process,” Jiao stated in a press briefing covered by the LA Times.
The company is releasing a free Life Sciences research plugin connecting to over 50 scientific databases and tools. Enterprise users who gain access to GPT-Rosalind will combine this reasoning layer with the plugin’s capabilities.
Launch partners for the model include Amgen, Moderna, and Thermo Fisher Scientific. OpenAI also has a separate research collaboration with Los Alamos National Laboratory on AI-guided protein and catalyst design.
Access to GPT-Rosalind is strictly limited to qualified U.S. enterprise customers following a safety review. This restricted rollout is a direct response to rising biosecurity concerns, including calls from scientists for tighter controls on biological AI training data.
No fully AI-discovered drug has yet cleared phase 3 clinical trials, as noted in recent reporting. The model’s potential impact hinges on its ability to compound small time savings across thousands of research labs.
