OpenAI Launches GPT-Rosalind for Drug Discovery

OpenAI releases GPT-Rosalind, a specialised AI model for life sciences research, drug discovery, and genomics. Available to select enterprise partners.

AI Tutorials · · 2 min read

Quick answer

OpenAI released GPT-Rosalind on April 16, 2026 — its first AI model built specifically for life sciences. It outperforms GPT-5.4 on biology benchmarks, can design experiments, query scientific databases, and parse research papers. Access is gated to enterprise partners including Amgen, Moderna, and the Allen Institute.

OpenAI has released GPT-Rosalind, its first AI model built specifically for life sciences research. Named after Rosalind Franklin, the scientist whose X-ray work was critical to discovering DNA’s structure, the model is designed to accelerate drug discovery, genomics, and translational medicine.

What Makes It Different

Unlike ChatGPT’s general-purpose approach, GPT-Rosalind was trained on 50 specific biological workflows. It can query specialised databases, parse recent scientific papers, interact with computational biology tools, and suggest new experimental pathways — all within a single interface.

Crucially, it’s designed to push back. Where a general model might enthusiastically endorse any hypothesis, OpenAI says GPT-Rosalind will reject poor drug targets and question weak experimental connections. That kind of critical reasoning is what researchers actually need.

On BixBench, a benchmark for real-world bioinformatics tasks, GPT-Rosalind scores 0.751, ahead of GPT-5.4 (0.732) and Gemini 3.1 Pro (0.550). On LABBench2, which covers protocol design and molecular cloning, it outperforms GPT-5.4 on six of eleven tasks.

Who Gets Access

This is not a consumer product. GPT-Rosalind is available through OpenAI’s trusted-access program, starting with US-based enterprise customers. The first partners are Amgen, Moderna, the Allen Institute, and Thermo Fisher Scientific.

OpenAI is also launching a Life Sciences plugin for Codex that connects to over 50 scientific tools and data sources, giving researchers programmatic access to biological databases through a developer interface.

What This Means for You

If you work in biotech or pharma, GPT-Rosalind could meaningfully speed up early-stage research — from literature reviews to experiment design. For everyone else, it signals a broader shift: AI companies are moving beyond general-purpose chatbots toward specialised models trained for specific professional domains.

This follows a busy week for AI model releases, including Anthropic’s Claude Opus 4.7 on the same day. To keep up with the latest AI tools, check out our newsletter.

Frequently asked questions

What is GPT-Rosalind?
GPT-Rosalind is OpenAI's first specialised AI model for life sciences research. Named after scientist Rosalind Franklin, it's trained on 50 biological workflows and designed for drug discovery, genomics, and translational medicine.
Can anyone use GPT-Rosalind?
Not yet. Access is limited to qualified enterprise customers through OpenAI's trusted-access program. Initial partners include Amgen, Moderna, the Allen Institute, and Thermo Fisher Scientific.
How does GPT-Rosalind compare to GPT-5.4?
On BixBench, a bioinformatics benchmark, GPT-Rosalind scores 0.751 compared to GPT-5.4's 0.732. It also outperforms GPT-5.4 on 6 out of 11 LABBench2 tasks covering areas like molecular cloning and protocol design.
What can GPT-Rosalind actually do?
It can synthesise evidence from scientific literature, generate hypotheses, design experiments, query biological databases, and evaluate drug targets — rejecting weak candidates rather than endorsing everything.

Want to keep learning?

Explore our guided learning paths or try building something with AI right now.

Enjoyed this article?

Subscribe for more AI insights delivered to your inbox every week.

No spam. Unsubscribe anytime.