Anthropic unveiled Claude Science on 30 June at an invite-only briefing in San Francisco for pharmaceutical executives, biotech founders, and academic researchers. The platform, now in public beta, is designed to do for scientific research what the company's Claude Code product has done for software engineering: replace a fragmented collection of specialist tools with a single, AI-driven environment.
One workspace, dozens of tools
The typical computational researcher's day involves constant switching between databases, coding notebooks, visualisation software, and high-performance computing terminals. Claude Science attempts to end that fragmentation. According to Anthropic, the platform connects to more than 60 scientific databases and ships with prebuilt toolkits covering genomics, protein structure, proteomics, and cheminformatics.
The architecture is built around a hierarchy of AI agents. A coordinating agent takes instructions in plain language and delegates subtasks to specialist sub-agents, while a separate reviewer agent checks every citation and calculation for accuracy before results are finalised. As MIT Technology Review reports, the platform also helps scientists run code on powerful computing clusters, which are essential for large-scale biological analysis but notoriously difficult to manage.
Crucially, Anthropic is not launching a new underlying model. As TechCrunch notes, Claude Science runs on the same Claude models already available to subscribers, including Claude Opus 4.8, with no special biological fine-tuning or gated access. The bet is on workflow, not on a more capable model.
Reproducibility built in from the start
Scientific reproducibility, the ability to independently verify a published result, has been a persistent problem in research for years. Claude Science takes direct aim at it. When the platform generates a figure or a data output, it saves the exact code and computing environment used to produce it, alongside a plain-language explanation and the full message history. Researchers can trace any result back to its source weeks or months later.
“"If Claude Science is able to automate so much of the information gathering … and help inform the ultimate decisions about where to go next, it would increase the pace of our experimentation by orders of magnitude." — Michael Pollastri, Northeastern University researcher”
Early adopters report significant time savings. At the UCSF Brain Tumor Center, epidemiologist Stephen Francis used Claude Science to carry out comprehensive germline analysis of glioma, a type of brain tumour, in roughly one-tenth of the time the work previously required, with results independently validated. At the Allen Institute, neuroscientist Jérôme Lecoq built a multi-agent computational review pipeline using the tool; according to Anthropic, comparable reviews had previously taken his team up to two years to complete.
A crowded field with competing bets
Claude Science enters a market that is rapidly filling with rival offerings. In April, OpenAI released GPT-Rosalind, a model fine-tuned specifically for biological reasoning. Google DeepMind, meanwhile, holds a structural advantage: its Gemini for Science platform bundles proprietary foundational models, including AlphaFold and AlphaGenome, alongside more than 30 life sciences databases. As TechCrunch observes, three distinct distribution strategies are now competing for the same scientific market — broad subscription access from Anthropic, an enterprise-gated approach from OpenAI, and proprietary model ownership from Google.
Anthropic's head of life sciences, Eric Kauderer-Abrams, positioned Claude Science as central to the company's core mission rather than a side project. Not all scientists are wholly convinced. Jared Auclair, a researcher at Northeastern University, told Northeastern Global News he is cautiously optimistic but warned that "a general-purpose AI can hallucinate or miss nuance in regulatory guidance or assay design," adding that the tool is "a co-pilot that requires a skilled pilot." The platform is available in beta on macOS and Linux for Pro, Max, Team, and Enterprise subscribers. Anthropic is also offering discounted team plans for academic labs and nonprofit research organisations, and will fund up to 50 external research projects with grants of up to $30,000 in platform credits. Applications are open until 15 July 2026.
“"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." — Eric Kauderer-Abrams, Anthropic head of life sciences”
This article is free to read. It always will be — no paywall, no account, no tracking.
