THE ESSENTIALS IN 30 SECONDS

  • Anthropic says 20 foundations, institutions and research groups now take part in AI for Science, launched in May 2025 to give researchers free Claude credits.
  • Partners include the US National Science Foundation, the Sloan and Simons foundations, the Wellcome Sanger Institute, the Francis Crick Institute and EMBL.
  • With HHMI Janelia, Anthropic released the Model Hardware Standard, a research-preview specification for letting AI agents operate lab instruments safely.

Anthropic announced on October 28 that its AI for Science program, which hands out free Claude credits to researchers, now works with 20 foundations, institutions and research groups. The program started in May 2025 and, according to the company, has so far backed hundreds of projects in biology, medicine, physics, chemistry, climate science and mathematics.

Who is joining

The new partners fall into three groups. The first is funders and scientific societies: the US National Science Foundation, the Alfred P. Sloan Foundation, the Simons Foundation, Schmidt Sciences, the Open Source for Science Fund run by Renaissance Philanthropy, and CAS, a division of the American Chemical Society. TUN reports that these organizations will pass credits on to the researchers they support.

The second group is research institutions: 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. HHMI's Janelia Research Campus and the Allen Institute, which Anthropic had already partnered with, are also counted; the Allen Institute is building teams of AI agents to analyze its data.

The third group covers specific projects, including the FlyWire brain-mapping effort at Princeton, Columbia's LEAP climate center, Johns Hopkins' AstroPath cancer imaging project, the Broad Institute's gnomAD genetic database, the Tau Ceti mathematics library built in Lean, and Australia's Centre for Population Genomics.

A standard for lab instruments

Alongside the expansion, Anthropic and HHMI Janelia published the Model Hardware Standard, a shared specification meant to let AI agents control laboratory instruments safely. It is labelled a research preview, and the announcement does not say which instruments or vendors already support it.

What is not stated

Anthropic's page does not give credit amounts for any partner. TUN reports that researchers outside partner organizations can still apply individually for up to $50,000 in credits per project. The expansion comes as rivals pursue similar efforts: TUN notes that OpenAI and Google have announced programs giving scientists and European researchers access to their models. How much the credits change research output, and how the hardware standard will be adopted, remains to be seen.