OpenAI's New Drug Discovery AI Beats 95% of Human Experts

A new artificial intelligence model could help bring medicines to patients years faster than the current 10 to 15 year timeline.

OpenAI unveiled GPT-Rosalind on April 16, 2026, marking the company's first AI built specifically for drug discovery and biological research. Unlike general chatbots, this specialized tool tackles the scientific heavy lifting that slows down early drug development.

The model excels at synthesizing thousands of research papers, connecting disconnected experimental results, and designing rigorous experiments. In tests with Dyno Therapeutics, GPT-Rosalind scored above the 95th percentile of human experts when predicting how certain DNA sequences would function.

Named after pioneering scientist Rosalind Franklin, whose work in the 1950s revealed DNA's structure, the AI reflects her data-driven precision. The model integrates with over 50 scientific databases and laboratory tools through a dedicated research plugin.

Major pharmaceutical and biotech companies are already partnering with OpenAI. Amgen, Moderna, the Allen Institute, and Thermo Fisher Scientific have signed on to use the technology in their research labs.

The bottleneck in drug development isn't just complexity. Scientists spend years manually reviewing literature, evaluating past hypotheses, and planning experiments. GPT-Rosalind automates these labor-intensive early stages, freeing researchers to explore more possibilities and surface hidden connections between studies.

The Ripple Effect

Faster drug discovery means patients with rare diseases could see treatments arrive years sooner. When development timelines shrink, pharmaceutical companies can pursue medicines for smaller patient populations that weren't previously economical to research.

The technology also makes cutting-edge research tools accessible to smaller labs and universities. Institutions without massive research budgets can now tap into the same AI capabilities used by major pharmaceutical corporations.

This shift toward specialized AI models could accelerate breakthroughs across multiple diseases simultaneously. While one team uses GPT-Rosalind for cancer research, another might apply it to develop antibiotics or design personalized gene therapies.

OpenAI is offering the model through ChatGPT, their Codex platform, and direct API access for qualified research organizations. The company views this as their entry into high-impact scientific domains where precision matters most.

The race to speed up biomedical discovery just got a powerful new tool, and researchers around the world are getting early access to see what becomes possible.