Anthropic unveiled Claude Science on Tuesday, positioning it as an AI‑powered workbench that consolidates the entire computational research workflow into a single environment. The company emphasizes that the service does not introduce a new model; it runs the same Claude models already available to customers, such as Claude Opus 4.8, with no special gating or extra access requirements.
The workbench builds on Anthropic’s October 2025 launch of Claude for Life Sciences, which enhanced the chatbot’s ability to handle biology‑related queries. Claude Science extends that capability by providing a dedicated space where scientists can manage data, run analyses, generate figures and keep a complete audit trail without hopping between disparate tools.
At the core of the platform is a single AI assistant that functions like a project manager. It links to more than 60 scientific databases and ships with prebuilt toolkits for fields ranging from genomics to protein‑structure prediction and chemistry. Users can spin up sub‑assistants to handle specific tasks, or hand work off to custom “expert” assistants they have built themselves. A separate fact‑checking AI reviews citations and calculations before any output reaches publication, aiming to curb the rise of fabricated references that have plagued AI‑assisted writing.
Reproducibility is baked into Claude Science. When the system generates a figure—say, a 3D protein model—it also records the exact code, environment and a plain‑language description of how the image was produced. Researchers can edit figures by describing desired changes in natural language, prompting the assistant to modify the underlying code automatically.
Another practical advantage is the option to run the workbench on a lab’s own infrastructure, keeping data in‑house rather than sending it to Anthropic’s servers. Early users have already put the platform to the test. Jérôme Lecoq, a neuroscientist at the Allen Institute, used Claude Science to build a multi‑agent computational review pipeline. Stephen Francis’s team at the UCSF Brain Tumor Center accelerated a comprehensive germline analysis of glioma, cutting the time required to a fraction of the previous effort while still obtaining independent validation.
The launch arrives amid competing approaches from other AI leaders. OpenAI introduced GPT‑Rosalind in April, a specialized model fine‑tuned for biological reasoning that is currently limited to qualified enterprise customers in the United States. Google DeepMind, meanwhile, leverages its proprietary AlphaFold and AlphaGenome models within a Gemini for Science platform that bundles over 30 life‑science databases.
Anthropic’s strategy differs by offering Claude Science through a broad subscription model. The service is in beta for anyone on Pro, Max, Team or Enterprise plans, and the company has announced up to $30,000 in credits for up to 50 projects that explore the frontiers of biomedical research. Applications are open through July 15, 2026, with awards announced by the end of that month and projects running from September to December 2026. Novo Nordisk and the Allen Institute were named as early customer case studies, suggesting that major pharma firms are already evaluating multiple AI vendors.
Este artículo fue escrito con la asistencia de IA.
News Factory APP - noticias agénticas para impulsar tu SEO y AEO.