Anthropic announced on October 28, 2026, that its AI for Science program now includes partnerships with 20 scientific foundations, institutions and research groups. The expansion provides free Claude credits for high-impact research, with participants including the US National Science Foundation, Space Telescope Science Institute, Wellcome Sanger Institute, Francis Crick Institute and European Molecular Biology Laboratory.
The change broadens how researchers can receive support. Funders and scientific societies will distribute credits to researchers, while participating institutions will use credits within their own organizations. Anthropic says many projects will produce resources available to the wider scientific community, including open datasets, open-source software, open standards and shared libraries. The announcement is an access-and-partnership update, with the scientific results still to come.
The most consequential detail is that public-goods goal. Free model access benefits the researchers receiving it; a useful, openly available tool or dataset could keep benefiting other labs after the original credits have been spent.
How Free Claude Credits Reach Researchers
Anthropic launched AI for Science in May 2025, initially supporting individual scientists. The company says it has since supported hundreds of projects across biology, medicine, physics, chemistry, climate science and mathematics.
Under the expanded arrangement, organizations provide two routes into the program:
- Funders and societies distribute credits. These organizations can connect Claude access with researchers and projects they support.
- Research institutions use credits internally. Their scientists can apply the resources to work within an established research environment.
That distinction matters. A foundation can reach researchers across multiple universities or specialties. An institute can bring model access into an organization that already operates shared data resources, instruments and computational workflows.
Anthropic says scientists are using Claude to generate hypotheses, design experiments, analyze complex data and communicate findings. These are broad categories of work, spanning exploratory reasoning through analysis and writing. The announcement does not present a single standardized workflow for every participating organization. Anthropic’s program description leaves the research questions with scientists.
The company also mentions its discounted Team plan for scientists. That is a separate access measure: a discounted subscription and a grant of free credits have different implications for a lab’s budget.
For researchers considering participation, practical questions include the size and duration of an allocation, which models and interfaces it covers, and what happens when the credits run out. The headline count of 20 partners cannot answer those project-level planning questions.
Which Organizations Are Participating?
Anthropic names six funders and societies in its announcement: the Alfred P. Sloan Foundation, Simons Foundation, Schmidt Sciences, Open Source for Science Fund at Renaissance Philanthropy, US National Science Foundation, and CAS, a division of the American Chemical Society.
Their roles differ. The foundations support research across various scientific disciplines. The Open Source for Science Fund focuses on the software stack underpinning science, while CAS maintains chemistry information resources, including the CAS REGISTRY system for identifying chemical substances.
The institutional partners named in the available announcement include:
| Institution | Research work or infrastructure highlighted |
|---|---|
| Space Telescope Science Institute | Science operations for the Hubble, James Webb and Roman space telescopes, alongside a major astronomy archive |
| Wellcome Sanger Institute | Large-scale genomic and functional data, spanning human genetics, biodiversity and drug development |
| Francis Crick Institute | Biomedical research into the biology behind health and disease |
| Imperial College London, Department of Life Sciences | Research from molecular structures to ecosystems, including health and climate resilience |
| European Molecular Biology Laboratory, or EMBL | Experimental and computational biology, plus shared services and open data resources |
| Gladstone Institutes | Biomedical research integrating AI and data science |
These are selected participants from the 20 organizations Anthropic highlights, not a complete roster. The descriptions come from Anthropic’s announcement.
The institutional route could be particularly useful where a research task depends on shared infrastructure. EMBL’s European Bioinformatics Institute, for example, hosts resources such as Ensembl, Protein Data Bank in Europe and Europe PMC. Connecting AI-supported work to established resources offers a plausible route to reusable scientific tools, though each project still needs to demonstrate its value.
Sanger Confirms Its Planned Use
The Wellcome Sanger Institute published its own announcement on October 29, 2026, confirming participation. It says its scientists will use Anthropic’s AI-powered research platforms to automate and accelerate complex computational biology and genomics analysis.
That provides a partner-side account of the intended work. It also gives the collaboration a more specific focus than general-purpose AI assistance: computational analysis in a genomics research setting. Sanger’s statement describes planned use, without reporting a measured improvement from this partnership.
Two Infrastructure Projects Show the Direction
Anthropic points to earlier collaborations with HHMI Janelia Research Campus and the Allen Institute as examples of its institutional approach. They address different parts of the research process: interacting with physical laboratory equipment and analyzing scientific data.
A Shared Specification for Lab Instruments
Anthropic says it worked with HHMI Janelia Research Campus to develop the Model Hardware Standard, or MHS. The company describes MHS as a shared specification intended to let AI agents safely operate lab instruments. It remains in research preview.
The useful idea is a common interface between software agents and scientific equipment. In principle, shared specifications can reduce the amount of custom integration needed when different software systems interact with instruments. A reusable interface could therefore matter beyond one experiment or one model.
The research-preview status is worth noting, though. Anthropic’s short description is not enough to assess MHS’s technical safeguards or its suitability for a particular instrument. I’d treat it as an infrastructure project to watch, with deployment decisions depending on the specification and equipment-specific validation. .
Frequently Asked Questions
3 questions
1What Does AI for Science Provide?
The program provides free Claude credits for high-impact scientific research. The expansion distributes access through participating funders, societies and research institutions.
