Anthropic opens a call for life sciences researchers: 8–12 weeks with Claude, free of charge

Anthropic has launched an open call for guest researchers in life sciences. Lab heads and their groups at universities, nonprofit organizations and government institutes can apply for a joint project with the company's researchers. The format: 8 to 12 weeks of intensive work, with Claude models and harnesses run at Anthropic's expense, and results published together with open data and code. Applications are accepted for roughly 14 days, and a reply arrives by email about 14 days after the call closes.

The program focuses on five areas: causal human biology, structural and computational molecular biology, medicinal chemistry, drug discovery and clinical science. That covers the full chain from understanding a disease mechanism to testing a hypothesis in patients.

What labs get and what Anthropic keeps

A guest researcher gets more than model access. They work side by side with people who know how Claude behaves on long, multi-step tasks, where it makes mistakes, and how to build a working pipeline around it. In essence, Anthropic is offering a science team for hire, bundled with a compute budget.

The intellectual property terms are written firmly in the scientists' favor:

• project length: 8–12 weeks, with extensions possible for some projects;
• cost of running models for participants: zero;
• data rights: stay with the researcher at every stage;
• rights to follow-on independent research: also theirs;
• results: published jointly, with data and code released as open source;
• who can apply: group leaders from academic institutes, nonprofits and government labs.

Anthropic stresses separately that it does not ask applicants to hand over data, either in the application or during the work. Only public information is allowed in application materials. In biomedicine, where closed patient datasets and unpublished screens are expensive to produce, this removes the main barrier to participation.

Why the 8–12-week window is the main constraint

The company openly admits that most open problems in biology will take years to resolve. But it will select projects that can deliver at least preliminary evidence within two to three months: an initial discovery, an answer to a question that has hung in the literature, an open dataset, or a public resource for the community.

That framing filters out vague pitches like "we'll speed up drug development with AI." Applicants need to explain exactly what result will appear in which weeks, and why it requires direct work with Anthropic's researchers rather than just an API key. The second point is critical: applicants must show that they benefit from the collaboration, not from free tokens.

The context for the program is clear. Anthropic's site has a nearby post describing how Claude discovered a new enzyme system with CRISPR-like repeats. The company appears to want to turn isolated stories into a steady flow: take on dozens of problems, see where the model actually moves science forward, and collect reproducible examples.

It is also part of a broader line of programs. Scientists who need direct access to the most powerful public models for biology can apply to the Life Sciences Verification Program. The AI for Science initiative already hands out Claude credits to researchers. The guest program fills a third niche: not access and not money, but joint work on a problem.

How to apply and what to prepare

The window is short for labs, so it is better to assemble a draft in advance. What makes sense to prepare:

• one clearly defined problem with a verifiable intermediate result within 8–12 weeks;
• a description of which Anthropic skills and tools fill a gap in your group;
• a list of artifacts you are ready to release publicly: data, code, annotations;
• application text built only on public data, with no closed datasets or unpublished results;
• a plan in case the project asks for an extension.

The signal for the industry is clear too. AI labs are increasingly moving from a "we sell a model" mode to a "we enter the scientific process" mode. Whoever first accumulates a body of confirmed discoveries and open resources will have a stronger argument than any benchmark. For pharma and biotech, it is also a chance to see results in the open before building their own pipelines around Claude.

Anthropic promises to expand its efforts if the program shows how best to support the scientific community. The first cohort and its publications will be the test: if 12 weeks produce data that can be reproduced, the format will scale quickly.