Anthropic launches Claude Science platform to boost scientific research

Anthropic announced the launch of Claude Science, an AI‑driven platform that unifies access to more than 80 scientific databases and a suite of analysis tools. The system acts as a project‑manager assistant, helping researchers review literature, plan experiments, analyse data, generate figures and manuscripts, and verify citations before publication. Early testers, including the Allen Institute and UCSF Brain Tumor Center, reported faster workflow and validated results. The company also said the platform will support pre‑clinical drug development for neglected diseases.

At the same time Anthropic unveiled Claude Sonnet 5, a cost‑reduced, more autonomous model that can plan actions, use browsers and terminals, and execute complex digital workflows with less human oversight. Pricing starts at $2 per million input tokens and $10 per million output tokens. Additionally, the firm released Claude Fable 5 for Claude Code, a model designed for asynchronous, long‑duration programming tasks, allowing developers to delegate complex code‑base work and maintain context over hours or days. These releases extend Anthropic’s effort to provide safer, more controllable AI across scientific and software development domains.