Ten thousand scientists get Claude free for a year

On Thursday, August 27, Anthropic announced a team plan just for scientists. Anyone who clears verification — a principal investigator or equivalent, at a university or nonprofit lab — brings the lab along. A standard seat, free, for a year. A premium seat, with five times the usage cap, at 15 dollars a month. The company says it intends to go well past those 10,000 seats in the coming months.

The cut is not “AI replaces the researcher.” It is access. In June the company launched Claude Science, a workbench with the tools a lab already uses, auditable artifacts and on-demand compute. The AI for Science program, until now concentrated on biology, is opening credits to other fields — including heavy compute research, of the kind the company cites in work on the Riemann zeta function and in protein design. Labs that outgrow the plan can apply for up to 50,000 dollars in credits per project. Any researcher can enter that second funnel.

There is a limit Anthropic writes in large type. People working in biology and chemistry stay on the Opus floor. Fable models still block professional biology and drug-development queries, because of dual-use risk. For the floor above, Mythos, the company says it is building an access program with the United States government, and that the first participants are already in. More people, later.

The announcement landed on the same day as the Model Hardware Standard, the spec for an agent to drive bench equipment. Those are different cuts. One is the badge: who may talk to the model. The other is the bench: what the model may switch on. Together they sketch the week Anthropic tried to occupy the lab from both sides.

For a reader who only wants to know what is going on, the pointer is this. Ten thousand scientists get a year of Claude. Biology still has a ceiling. The microscope is a different story.