OpenAI Launches GPT-Rosalind, a Research AI Model for Biology and Drug Discovery

OpenAI Launches GPT-Rosalind for Life Sciences Research

OpenAI introduced GPT-Rosalind on April 16, 2026. The model is named after Rosalind Franklin, the British scientist who helped discover the structure of DNA. It is OpenAI's first model built specifically for life sciences work.

What It Does

GPT-Rosalind acts as a specialized reasoning layer for scientific work. It reads published evidence, reviews experimental data, uses lab tools, and helps plan experiments. It is fine-tuned for genomics, protein engineering, and chemistry.

Typical tasks include synthesizing findings across hundreds of papers, generating biological hypotheses, and drafting experiment plans. These are jobs that usually take senior researchers weeks of literature review and discussion.

Performance

On BixBench, a benchmark for real-world bioinformatics and data analysis, GPT-Rosalind leads among models with published scores. On LABBench2, a more detailed test, it outperformed GPT-5.4 on six of eleven tasks. The biggest gains showed up in CloningQA, a test of molecular cloning reasoning.

Access

OpenAI is not releasing the model openly or to the general public. It is launching through a Trusted Access program for qualified enterprise customers in the United States. Early partners include Amgen, Moderna, the Allen Institute, and Thermo Fisher Scientific. OpenAI also expanded its Codex plugin on GitHub with connections to over 50 science tools and data sources.

Why It Matters

General models like GPT-5.4 can handle science questions, but they miss the fine detail that working scientists need. GPT-Rosalind is a signal that OpenAI wants domain-specific models to sit next to its general ones. If the Trusted Access pilot works, expect similar specialized models for other fields like materials science and clinical medicine.