OpenAI Ships GPT-Rosalind, a Specialist Model for Drug Discovery and Life Sciences

OpenAI shipped GPT-Rosalind on April 16, a specialist reasoning model for life sciences research. The model is named after British chemist Rosalind Franklin, whose X-ray crystallography data contributed to the discovery of DNA’s structure.

GPT-Rosalind targets the early stages of drug discovery. Claimed capabilities include evidence synthesis across research corpora, hypothesis generation from existing literature, and multi-step experimental planning. On the BixBench biomedical benchmark OpenAI reports leading performance, and the model outperforms GPT-5.4 on 6 of 11 tasks on LABBench2.

Access is not general. The initial rollout is a research preview gated through a trusted-access program for qualified enterprise customers in the United States. Anchor customers include Amgen (biotech), Moderna (vaccines), and the Allen Institute (nonprofit bioscience). OpenAI has built in safety controls including systems to flag potentially dangerous use and technical limits on what the model can produce without authorisation.

Available surfaces are ChatGPT, Codex, and the OpenAI API. Pricing and rate limits have not been publicly disclosed.

The release places OpenAI in direct competition with Google’s AlphaFold spin-offs and Anthropic’s Project Glasswing-linked initiatives for frontier vertical-specialty models. It is also OpenAI’s clearest statement yet that the general GPT line will branch into task-specific frontier models rather than absorbing every use case into one system.

Sources

• OpenAI: Introducing GPT-Rosalind
• Bloomberg: OpenAI Takes on Google With New AI Model Aimed at Drug Discovery
• Fierce Biotech: OpenAI launches biotech-specific AI model, GPT-Rosalind