The company said it wanted to remove the tedious procedural aspects inherent to scientific research by uniting fragmented tools, resources, file formats and databases.

Anthropic, the AI company behind Claude and Mythos, has unveiled its ‘Claude Science’ offering, which it described as “an AI workbench for scientists”.

Classified as a “public beta app” that runs using existing Claude models, Anthropic said its new product would “integrate the tools and packages that researchers most commonly use” in order to produce “auditable artifacts”, and provide “flexible access to computing resources”.

The company said it aimed to remove the tedious procedural aspects inherent to scientific research by uniting fragmented tools, resources, file formats and databases “into a single research environment where scientists can conduct all stages of their work”.

The app, which is now available in beta via Claude Pro, Max, Team and Enterprise plans, can help scientific users analyse literature, execute multi-step research, produce detailed artifacts, and iteratively refine figures and manuscripts prior to publication, according to its maker.

Anthropic, in a blogpost detailing the release, said that Claude Science can natively render “rich scientific artifacts, including 3D protein structures, genome browser tracks, chemical structures and more”, alongside generated plain-language descriptions of how such figures were created, to be used for later validation, record-keeping and reproduction.

The app can also handle planning and resource allocation for large-scale analyses that would typically require separate monitoring and computing capacity decisions to be made by a researcher or team, according to Anthropic.

“As the pipeline runs, a reviewer agent inspects the outputs, flagging incorrect citations, untraceable numbers and figures that don’t match their underlying code, and self-correcting as it goes,” the blogpost read.

The app is also said to be capable of synthesising answers to user questions through consultation of a wide range of databases and trusted sources of scientific information, which can be customised to user preferences.

Anthropic noted that in recent months, researchers have used Claude Science in beta for tasks such as single-cell RNA sequencing analysis, protein structure prediction, cheminformatics and more.

Meanwhile, Bloomberg reported that Anthropic has started to work on in-house, preclinical drug discovery schemes outside of the traditional scope of biotech and pharma research.

Earlier this week, Google Cloud Marketplace said it would begin offering two ‘large quantitative models’ (LQMs) developed by SandboxAQ later in 2026 with the aim of driving AI-assisted developments in materials science, healthcare and drug discovery.

SandboxAQ is already integrated with Anthropic’s Claude AI model. It claims its LQMs can offer “critical advances” in sectors such as life sciences, financial services and navigation.

In other Anthropic news, after weeks of uncertainty around the status and availability of Claude Fable 5 and Mythos 5 due to an impasse between the US government and Anthropic, the AI models had their export bans lifted by the country’s Department of Commerce yesterday (30 June).

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.