Anthropic Debuts Claude Science and Launches Internal Drug Discovery Initiative for Neglected Diseases

A New Workbench for Scientific Research

On June 30, 2026, Anthropic launched Claude Science, a specialized AI research workbench designed to assist scientists and drugmakers. The platform aims to accelerate early-stage drug discovery by helping researchers with experiment planning, literature analysis, scientific reasoning, and biological workflows. By offering these capabilities, the company is betting that its generalist Claude models will find widespread adoption across the pharmaceutical industry.

Direct Entry into Preclinical Drug Development

In a significant strategic expansion, Anthropic is not merely providing software to external pharmaceutical companies but is also launching its own preclinical drug discovery program. This internal initiative will focus on neglected diseases, which have historically struggled to attract traditional funding and research due to their limited commercial markets. Anthropic executives indicated that engaging in firsthand drug development using Claude Science will yield critical insights to refine and improve their AI tools.

Building on a Growing Healthcare Ecosystem

The launch of Claude Science builds upon a series of medical and life sciences updates Anthropic introduced earlier in the year. On January 11, 2026, the company announced Claude for Healthcare, a suite of HIPAA-ready products designed for providers, payers, and health technology startups. This followed the October release of Claude for Life Sciences, which integrated the AI into scientific platforms to manage clinical trials and regulatory operations.

Underpinned by Advanced Model Intelligence

These specialized scientific tools leverage the capabilities of Anthropic's latest model, Claude Opus 4.5. When evaluated using native tool use and extended thinking of 64k tokens, the model demonstrated strong performance on complex biomedical simulations. These include SpatialBench, a spatial biology analysis benchmark created by LatchBio, MedCalc for medical calculation accuracy using Python code execution, and MedAgentBench, a medical agent task completion benchmark developed by Stanford.

By directly entering the high-risk, low-margin arena of neglected disease research, Anthropic is testing whether AI-driven efficiency can fundamentally rewrite the economic realities of pharmaceutical development.