Claude Science

Claude Science is Anthropic’s scientific workbench, launched on 30 June 2026 and aimed at computational biologists, bioinformaticians and life-science labs. Rather than wrapping a chat window around a model, it unifies scientific databases, code execution, HPC compute and manuscript drafting inside one research environment.

What it does

The workbench connects a reasoning model to the things a computational scientist actually touches: reference databases, a runnable code environment, cluster compute, and the draft manuscript. The intended loop is that a question becomes a query, the query becomes analysis code, the code runs against real data, and the result lands somewhere citable — without the researcher shuttling context between six tools by hand.

Strengths

• Grounded in connected databases rather than model memory alone.
• Executes analysis code instead of only describing it.
• Spans discovery through drafting in one environment.
• Built for computational biology specifically, not general knowledge work.

Limitations

It is new, it is a subscription product, and the deep integration that makes it useful also makes it a dependency. Output still requires the same verification any analysis does — a plausible pipeline that runs cleanly can still be wrong. Our comparison against the open-source alternative is in OpenScience vs. Claude Science.