GPT-Rosalind: OpenAI Builds Its First AI Model Dedicated to Drug Discovery and Life Sciences

Getting a new drug from lab bench to pharmacy shelf takes 10 to 15 years in the United States, and only 1 in 10 candidates survive clinical trials. OpenAI thinks AI can compress that timeline, and it just shipped its first purpose-built model to prove it.

What is GPT-Rosalind?

Announced on April 16, GPT-Rosalind is a frontier reasoning model fine-tuned for biochemistry, genomics, and protein engineering. Named after Rosalind Franklin, the chemist whose X-ray crystallography helped reveal DNA's double helix, the model is built to synthesize evidence, generate hypotheses, and plan experiments across multi-step scientific workflows.

OpenAI is also releasing a free Life Sciences research plugin for Codex that connects researchers to over 50 scientific tools and databases. That plugin works with the standard API models too, so you don't need Rosalind access to benefit.

GPT-Rosalind benchmark results

On BixBench, a bioinformatics benchmark, GPT-Rosalind scored 0.751 on Pass@1, beating GPT-5.4 (0.732) and Gemini 3.1 Pro (0.550). On LABBench2, it outperformed GPT-5.4 on 6 of 11 research tasks, with its strongest gain in CloningQA, which tests end-to-end molecular cloning protocol design.

In a partnership with Dyno Therapeutics using unpublished RNA sequences, the model's best submissions ranked above the 95th percentile of human experts on prediction tasks. Strong numbers, though independent validation in real lab settings will be the true test.

Controlled access and big pharma partners

You can't just sign up. GPT-Rosalind is available as a research preview in ChatGPT, Codex, and the API, but only for qualified U.S. enterprise customers through a Trusted Access program. Launch partners include Amgen, Moderna, the Allen Institute, and Thermo Fisher Scientific.

The restricted rollout is deliberate. A model that reasons about pathogens and protein engineering carries real biosecurity risks. OpenAI has built in monitoring systems that flag sensitive activity, and during the preview, usage won't consume existing credits.

Where GPT-Rosalind fits in the AI drug discovery race

OpenAI isn't entering an empty field. Google's Isomorphic Labs already has IsoDDE, described by scientists as an "AlphaFold 4," with pharma deals worth billions at Eli Lilly and Novartis. Amazon launched Bio Discovery for antibody pipelines. NVIDIA is pushing BioNeMo.

The strategic angle is different, though. DeepMind builds molecular prediction engines. OpenAI is betting on domain-specific reasoning across broader scientific workflows. Both approaches have merit, and researchers will likely use both. The real winner here might just be patients.

For more details on GPT-Rosalind and its early life sciences use cases, you can read OpenAI’s official announcement here: official announcement.