What is Claude Science? Anthropic's new AI tool for scientists, explained

If you saw that Anthropic launched 'Claude Science' and weren't sure what that means for anyone who isn't a scientist, here's the simple version. On 30 June 2026, Anthropic — the company behind the Claude chatbot — released a tool built specifically for research scientists, especially in biology and medicine. Instead of being a general chatbot, it's a workbench: one screen where a researcher can do most of the different jobs their work involves, without constantly switching between separate apps.

Picture a typical day for a life-sciences researcher. They search for papers in a database called PubMed. They write and run code in a Jupyter notebook or in R to analyse data. They log into a remote cluster terminal to run bigger computations. Each of those lives in a different place, and moving between them — copying results here, reformatting data there — eats hours. Claude Science's whole idea is to put all of that in one environment, with the AI able to reach into more than 60 scientific databases, use computing tools, and keep track of the workflow, so the researcher stays in a single place.

Right now it's a beta (an early version still being tested) and it's available to people on the paid Claude Pro, Max, Team and Enterprise plans, on macOS and Linux computers. Anthropic also set up a grants programme to encourage scientists to try it: up to 50 projects can each receive as much as $30,000 in Claude credits — essentially free usage — and a compute company called Modal is adding up to $2,000 of computing power for some projects. If you want in, applications are open through 15 July 2026, winners are chosen by the end of July, and funded projects run from September to December 2026.

This wasn't a sudden idea. Anthropic's boss, Dario Amodei, has said he wants AI to make life-sciences research about 10 times faster. To get there, Anthropic spent roughly $400 million buying a computational-biology startup called Coefficient Bio, and it hired John Jumper — a scientist who won a Nobel Prize in 2024 for building AlphaFold, a famous AI that predicts protein shapes. Claude Science is how all of that turns into something a scientist can actually use. And Anthropic isn't alone: OpenAI released its own science-focused model, GPT-Rosalind, back in April 2026, and Google DeepMind is also in the race.

One detail worth knowing if you follow this space: some of the biggest customers are trying more than one of these tools at the same time. The drugmaker Novo Nordisk, for example, is a named partner for both Anthropic and OpenAI, and the Allen Institute has worked with both as well. So even though the headlines make it sound like a winner-takes-all fight, in reality big research organisations are testing several AIs side by side before deciding — which means it's early days, and no single company has locked this up yet. If you are a researcher, the practical takeaway is simple: it is worth trying, especially with the free credits on offer, but you are not locked in — you can test Claude Science alongside OpenAI or Google and keep whichever actually saves you time. And if you are just curious, the bigger signal is this: the AI labs now see science, not just chat, as a prize worth fighting hard for.