What happened
On 28 October 2026, Anthropic announced an expansion of its AI for Science program, which provides free Claude credits to researchers. The program began in May 2025, and Anthropic says it has supported “hundreds of projects across biology, medicine, physics, chemistry, climate science, and mathematics”. The new phase extends it from individual scientists to foundations, institutions and research groups.
Disclosure: Anthropic makes Claude, the AI tool used to research and draft this article, so Anthropic’s own announcement is covered here with that conflict of interest in mind.
Key details
- Partners named by Anthropic: science funders and societies including the Alfred P. Sloan Foundation, the Simons Foundation, Schmidt Sciences and the US National Science Foundation; institutions including the Space Telescope Science Institute, the Wellcome Sanger Institute, the Francis Crick Institute, EMBL and the Gladstone Institutes; and research initiatives including FlyWire at Princeton, AstroPath at Johns Hopkins and gnomAD at the Broad Institute.
- Lab instruments: Anthropic describes a partnership with the HHMI Janelia Research Campus on a Model Hardware Standard, in research preview, intended to let AI agents operate lab instruments safely.
- Agents for data analysis: Anthropic says the Allen Institute is building teams of AI agents to analyze its data, “turning months of manual analysis into hours”. That is Anthropic’s description; the post gives no measured results.
- Public goods: Anthropic says many projects are expected to produce open datasets, open-source software and shared libraries.
- Stage: “Many of these collaborations are still in their early stages,” the post says, and it does not report outcomes for the new partners.
Why it matters
For researchers, the practical news is access: the program is a route to free Claude credits, now with institutional partners that may distribute them. The post does not give eligibility rules or credit amounts in the parts we reviewed, so check Anthropic’s application page before you plan around it.
The lab-instrument preview is the part to watch. It is a proposed standard, not a product with published results, so safety and reliability claims will need independent testing. Until the collaborations report outcomes, treat the announcement as a statement of intent from the vendor.
Sources
- AI for Science expands to foundations and research institutions Primary source , Anthropic