AI · 30 October 2026 · 4 min read

Anthropic is giving Claude credits to 20 science partners. The instrument standard is a preview

The company’s post names foundations, institutes and research groups, and says what they receive is free Claude credits. With the Janelia campus it says it is developing a Model Hardware Standard so agents can operate laboratory instruments. That standard, Anthropic says, is in research preview. A specification is not an instrument that has run an experiment.

A collage: pencil drawings of cells, a person at a microscope, and a blue-tinted photograph of a large telescope inside a dome. — Anthropic’s illustration for the science post: a notebook of cells, a microscope and a telescope. Not a laboratory running Claude, and not the instrument standard.

Anthropic has posted an expansion of its AI for Science programme. The post says the company is highlighting 20 scientific foundations, institutions and research groups it has partnered with, and that the programme gives free Claude credits to researchers on high-impact projects. A credit is use of the model. It is not a grant in pounds or dollars, and the post does not say how many credits each partner receives, or what happens when they run out.

The names are more checkable than the size of the credit. Among the funders and societies Anthropic lists are the Alfred P. Sloan Foundation, the Simons Foundation, Schmidt Sciences, the Open Source for Science Fund, the US National Science Foundation and CAS, a division of the American Chemical Society. Among the institutions are the Space Telescope Science Institute, the Wellcome Sanger Institute, the Francis Crick Institute, Imperial College London’s Department of Life Sciences, EMBL and the Gladstone Institutes. A named institute is a partner on a web page. It is not, by itself, a paper that used the model.

The line that reaches a machine is the Model Hardware Standard. Anthropic says it partnered with the HHMI Janelia Research Campus to develop that standard, a shared specification that lets AI agents safely operate laboratory instruments, and that the standard is currently in research preview. It also says it supports the Allen Institute. A specification in preview is a document someone can read. It is not a count of instruments an agent has been allowed to move, and the post does not name a laboratory where the standard is already in daily use.

The rest of the post is a map of fields, not a result. The projects it describes include FlyWire at Princeton, AstroPath at Johns Hopkins, the Broad Institute’s gnomAD and Columbia’s LEAP. Those are real research efforts. Anthropic’s sentence is that they are among the partners receiving credits. It does not say Claude has folded a protein, called a variant or pointed a telescope. Until one of those groups publishes what the model did that they were not already doing, the expansion is distribution of credits. It is not evidence that an instrument has been handed to an agent.