Introducing GPT-Rosalind for life sciences research

OpenAI’s first frontier model purpose-built for life sciences research.
- Targets biochemistry, genomics, protein engineering, translational medicine.
Named after Rosalind Franklin, key contributor to DNA structure discovery.
Drug dev bottleneck: 10–15 years target-to-approval; model attacks analytical grind.
- Literature synthesis, hypothesis gen, experimental design, DB queries — unified interface.
Codex Life Sciences plugin ships simultaneously: 50+ scientific tools/databases.
Benchmarks: 95th percentile RNA prediction, 84th percentile sequence generation (Dyno).
- BixBench 0.751 pass rate; outperforms GPT-5.4 on 6/11 LABBench2 tasks.
- Strongest gains in CloningQA (molecular cloning protocol design).
Gated US-only enterprise preview; partners: Amgen, Moderna, Allen Institute, Thermo Fisher.
- Los Alamos collaboration on protein and catalyst design.
- Free during preview; orgs must pass safety/governance qualification review.

X discourse

@kevinweil: “Super excited to launch GPT-Rosalind, our first frontier model built for scientific research across biology, drug discov” (900 likes)
@kimmonismus: “GPT-Rosalind is competent rather than revolutionary… confirms frontier labs fragmenting vertically, building industry-“ (376 likes)
@WesRoth: “GPT-Rosalind… optimized for complex scientific workflows in protein/chemical reasoning, genomics analysis.” (50 likes)
@Marktechpost: “OpenAI dropped GPT-Rosalind… accelerate drug discovery by reasoning through biochemistry and experimental design.” (20 likes)
@AllenInstitute: “Proud to be a launch partner for OpenAI’s GPT-Rosalind, a new model designed to support life sciences research.” (17 likes)
@thoughtson_tech: “GPT-Rosalind dropped… gated behind trusted access, free during preview, pharma names lined up (Amgen, Moderna).” (0 likes)

Type	Link
Added	Apr 20, 2026