Anthropic Opens 10,000 Free Claude Seats for Scientists

Anthropic on Thursday opened 10,000 Claude subscription seats for scientists worldwide, offering free standard access and discounted premium tiers through a new Claude Team plan for researchers. The announcement also expands the company’s AI for Science program — previously focused on biological sciences — to cover compute-intensive fields including mathematics, physics and engineering.

The timing is not accidental. Anthropic is entering a direct competition with OpenAI for the loyalty of the academic research community, and this program is its most concrete move yet to build that relationship at scale.

What Anthropic Is Actually Offering

The new Claude team plan for scientists runs for one year and has two tiers. Standard seats are free; premium seats — which carry five times the normal usage limits — cost $15 per month. Anthropic says it plans to grow the program well beyond the initial 10,000 seats in the coming months.

Access is gated at the principal investigator level. To qualify, an applicant must be a PI or equivalent at an academic or nonprofit research institution, complete a verification form, and then add the members of their lab to the plan. That structure matters for graduate students and postdocs: your access depends entirely on whether your advisor signs up.

Separate from the subscription seats, the AI for Science program offers up to $50,000 in compute credits per qualifying project, and any researcher — not just PIs — may apply. That grant track is the more direct route for early-career researchers to compete on merit rather than waiting on an advisor.

Anthropic also pointed to  Claude Science a workbench product launched in June 2026, which integrates tools researchers commonly use, generates auditable research artifacts, and provides flexible access to computing resources. It is the workflow layer sitting on top of the subscription access.

Model Access Has Real Limits — Especially in Biology

Not all researchers get equal access to Claude’s most capable models. Biologists and chemists are currently restricted to Opus-class models under the standard program. Anthropic’s more powerful Mythos-class models remain off-limits through this program because of their dual-use risk — the same capabilities that could accelerate drug discovery could theoretically be misused for harmful biological research.

Anthropic is working with the U.S. government on a separate, structured access program to give life sciences professionals controlled access to Mythos-class models. The company says it has enrolled its first participants and expects to expand that program soon. The partnership reflects a more cautious posture than competitors: Anthropic’s own research found that Mythos-class models can predict viral shell assembly better than specialized protein language models, which is precisely why access is being managed carefully.

Researchers in mathematics, computer science, physics and other newly included fields face fewer restrictions, which gives them a broader range of model capabilities from the start.

How This Stacks Up Against OpenAI

OpenAI launched its own program for academic researchers — ChatGPT for Academic Researchers — offering a year of access roughly equivalent to its $200-per-month ChatGPT Pro account, with up to five seats per approved workspace. OpenAI has set a target of reaching 100,000 researchers through 2027, a ceiling 10 times larger than Anthropic’s current rollout, though the programs are structured differently.

OpenAI’s offer is workspace-centric and institution-based; Anthropic’s is PI-managed and lab-based. OpenAI bundles access to GPT-5.6 Sol Pro, Codex, larger context windows and business-grade security. Anthropic’s premium tier is priced at just $15 per month — dramatically cheaper on paper — but comes with the model-access caveats described above, particularly for life sciences researchers who may find the Opus-class restriction a meaningful limitation.

What This Means for Students and Early-Career Researchers

For graduate students and postdocs, the most important thing to understand is the key-holder problem: the free seats are unlocked at the PI level, not the student level. If your advisor does not apply, you do not get in through this route. That makes it worth having a direct conversation with your PI about signing up, particularly if your lab does compute-heavy work in any of the newly covered fields.

For those in biology or chemistry, the Opus-class restriction is real but not necessarily disqualifying — Opus-class models are still highly capable, and the $50,000 grant track through AI for Science remains open to any researcher regardless of career stage.

More broadly, programs like this one shape which AI tools researchers reach for habitually. Students who build fluency in a particular ecosystem during graduate school tend to carry those preferences into their careers and labs. A free or near-free on-ramp to Claude — with an integrated research workbench — is a concrete way Anthropic is trying to make that ecosystem Claude.

Researchers interested in applying can find the verification form and program details here.

Source: Anthropic