Anthropic Launches Claude Science — an AI Workbench That Doesn't Need a New Model

Anthropic unveils Claude Science, an AI workbench purpose-built for scientific research. It's not a new model — it's the same Claude Opus 4, wrapped in a domain-specific operating layer that connects to 60+ scientific databases, runs on labs' own infrastructure, and is available now to all paid subscribers. The move positions Anthropic against OpenAI's GPT-Rosalind and Google DeepMind's AlphaFold in the race to own AI-for-science.

<p>On Tuesday, Anthropic introduced Claude Science — a dedicated AI workbench for scientific research that deliberately isn't a new model. It runs the same Claude Opus 4 already available, but wraps it in a purpose-built environment with direct connections to over 60 scientific databases, prebuilt toolkits for molecular biology and drug development, and the ability to generate publication-ready figures alongside their underlying code.</p>

<p>The thesis is simple: scientists don't need a smarter AI. They need one that knows where the data lives, can run reproducible computational pipelines, and won't send sensitive research data to a third-party cloud. Claude Science runs on the lab's own infrastructure. It can autonomously carry out multi-step research workflows — literature review, data analysis, figure generation, citation verification — when given concise, high-level instructions.</p>

<p>"It represents how important this is to our mission that this is right up there with Claude Code and Claude Cowork as the next really significant product that we're releasing," said Eric Kauderer-Abrams, Anthropic's product lead for science.</p>

<p>The launch, announced at an AI for Science briefing attended by pharmaceutical executives and biotech founders, marks Anthropic's second major push into the scientific domain. In October 2025, the company released Claude for Life Sciences — essentially plugins that made the chatbot better at domain-specific tasks. Claude Science is a full workbench, closer in spirit to Claude Code but built for the lab bench rather than the software IDE.</p>

<p>Early adopters are already putting it to work. Allen Institute neuroscientist Jérôme Lecoq used Claude Science to build a multi-agent computational review pipeline. Stephen Francis's group at UCSF deployed it for drug candidate screening. Anthropic named Novo Nordisk and the Allen Institute as case studies, and announced it will fund up to 50 academic projects with credits of up to $30,000 each — targeting postdoctoral and graduate work that "explores the boundaries of AI-driven scientific discovery."</p>

<p>The competitive landscape makes the launch especially interesting. OpenAI took a different path in April, releasing GPT-Rosalind — a specialized model fine-tuned for biology, available only as a research preview to select partners. Google DeepMind, meanwhile, owns the foundational models: AlphaFold and AlphaGenome, which both Anthropic and OpenAI can only call into as third-party APIs. Three fundamentally different distribution strategies are now competing for the same market: Anthropic going wide with broad subscription access, OpenAI going narrow with curated partners, and DeepMind going deep with proprietary foundation models.</p>

<p>Claude Science is available in beta to anyone on a Pro, Max, Team, or Enterprise plan. Anthropic says the company is on track for its first profitable quarter — and major contracts with pharmaceutical companies could help cement that trajectory.</p>

<p>Sources: TechCrunch, MIT Technology Review</p>