Enter your email address below and subscribe to our newsletter

OpenAI upgrades GPT-Rosalind with GPT-5.5 for drug discovery

Share your love

  • OpenAI upgraded GPT-Rosalind, its specialized life sciences model, on Tuesday with GPT-5.5’s agentic coding backbone and new drug discovery benchmarks.openai
  • The model, first launched in April, outperforms GPT-5.5 on chemistry, genomics, and wet lab tasks while using fewer tokens, according to OpenAI.openai
  • Partners including Novo Nordisk 0.76%, Amgen 1.19%, and Moderna 3.50% are applying the upgraded model across research workflows.openai

OpenAI Upgrades GPT-Rosalind With GPT-5.5 Backbone for Drug Discovery and Life Sciences

OpenAI on Tuesday announced a major update to GPT-Rosalind, its specialized life sciences reasoning model, combining the capabilities of GPT-5.5 with deeper intelligence in drug discovery, genomics, and medicinal chemistry. The upgrade marks the company’s most ambitious move yet to bridge frontier AI with practical scientific research workflows.x

Stronger Intelligence, New Benchmarks

The updated model brings GPT-5.5’s agentic coding and tool-use capabilities together with enhanced reasoning across core drug-discovery domains, according to OpenAI’s announcement on June 3. The company designed a new benchmark called LifeSciBench to evaluate real-world scientific performance across six workflow areas: evidence handling, analysis, design and optimization, scientific reasoning, validation and operations, and translation and communication.openai

On MedChemBench, which tests multimodal chemical structure understanding, structure-activity relationships, and lead optimization, GPT-Rosalind outperforms GPT-5.5 at 27.5% versus 25.1% while using fewer tokens. On GeneBench, which evaluates agentic performance on long-horizon quantitative biology tasks, the model achieves 21.6% accuracy compared to GPT-5.5’s 20.4%. A new LabWorkBench evaluation, testing the model’s ability to link perturbations to experimental outcomes in real wet lab protocols, shows GPT-Rosalind scoring 63.2% versus GPT-5.5 at 55.8%.openai

From Molecules to Living Systems

GPT-Rosalind was originally introduced in April as OpenAI’s first purpose-built model for life sciences, named after Rosalind Franklin, the chemist whose X-ray crystallography work was essential to revealing DNA’s structure. The model is designed to synthesize data across scales — from molecules to genes to pathways to living systems — and assist researchers with evidence synthesis, hypothesis generation, and experimental planning.openai

OpenAI also launched Life Sciences Research and Life Sciences NGS Analysis plugins for its Codex platform, extending the model with practical execution capabilities for bioinformatics workflows, sourced evidence retrieval, and biological interpretation.openai

Global Expansion and Partnerships

The company is expanding GPT-Rosalind access to eligible organizations globally through its trusted-access deployment structure. Novo Nordisk is among the partners leveraging the updated model. Mishal Patel, Group Vice President of AI and Digital Innovation at Novo Nordisk, said in OpenAI’s announcement: “To deliver meaningful value for researchers, advanced AI models must be grounded in trusted scientific data, connected to validated tools, and integrated into the real-world workflows researchers use every day.”openai

OpenAI is also working with Amgen, Moderna, the Allen Institute, and Thermo Fisher Scientific to apply the technology across discovery processes.pharmaphorum

Leave a Reply

Your email address will not be published. Required fields are marked *

Stay informed and not overwhelmed, subscribe now!