OpenAI upgrades GPT-Rosalind for life-sciences workflows

OpenAI's latest GPT-Rosalind update is aimed at making frontier models useful inside real scientific workflows, not just at answering biology questions. The company says the model combines GPT-5.5 agentic coding and tool use with stronger performance in medicinal chemistry, genomics, quantitative biology, evidence handling, and lab troubleshooting. OpenAI introduced LifeSciBench for expert-judged life-sciences tasks, MedChemBench for molecular work, GeneBench for long-horizon genomics analysis, and LabWorkBench for wet-lab protocol reasoning. Claimed gains include GPT-Rosalind scoring 27.5% versus GPT-5.5 at 25.1% on MedChemBench while using 7.2% fewer tokens, and 63.2% versus 55.8% on LabWorkBench. Access remains gated through a trusted-access research-preview structure, with Novo Nordisk named as a partner. This belongs in the feed because scientific AI is moving from benchmarks toward controlled enterprise research environments with provenance and safety constraints.

Key details: June 3, 2026, GPT-Rosalind, LifeSciBench, MedChemBench 27.5% vs GPT-5.5 25.1%, GeneBench 21.6% vs 20.4%, LabWorkBench 63.2% vs 55.8%, Novo Nordisk, Trusted-access research preview.

Continue swiping for more AI Brief stories.