Anthropic Gives 10,000 Scientists Free Claude Access, Lets AI Run Lab Hardware

Anthropic wants to be the operating system of American science. On Thursday, August 27, 2026, the company announced it's opening 10,000 free and discounted Claude Team subscription seats to scientists worldwide, according to Anthropic's own announcement covered by Unite.ai, Crypto Briefing, and TUN. Standard seats are free. Premium seats, with five times the usage limits, run $15 a month, locked in for a year.

Anthropic's published terms show the regular price for those premium seats is $100 a month. Free standard seats normally run $20. For a university lab running on grant money, that represents real savings.

There is a structural limitation, however. Access is gated at the principal investigator level, according to TUN. You have to be a PI, or equivalent, at an accredited academic or nonprofit institution, complete a verification form, then add your own lab members. If your advisor doesn't sign up, you don't get in through this door. For-profit companies and contract research shops are excluded entirely and pointed to standard commercial pricing.

There is a second, more open path. Anthropic's AI for Science program, running since May 5, 2025, is expanding beyond its original focus on biological sciences into math, physics, and engineering, and any researcher can apply for up to $50,000 in compute credits per project, not just PIs, according to TUN and nextaipress. That's the track early-career researchers without a well-connected advisor can actually use.

Not All Access Is Equal

Anthropic's most capable model tier, called Mythos, remains largely off-limits under this program. TUN reports biologists and chemists are currently restricted to Opus-class models specifically because of dual-use risk: the same tools that speed up drug discovery could theoretically help someone engineer something dangerous. A company handing frontier biology tools to any verified email address without restriction would be making a different bet on the same risks.

Separately, nextaipress reports Anthropic has enrolled the first participants in a government-linked access program giving life sciences researchers entry to Mythos-class models specifically, in partnership with the U.S. government. Anthropic did not name the agency involved or say how many participants have joined. This is a new, undisclosed arrangement, distinct from the export-control suspension of Anthropic's Fable and Mythos models tied to the Commerce Department in June 2026, which nextaipress says was about cybersecurity and national security, not biology research.

Claude Gets Hands

The subscription news landed alongside something more physically consequential. Anthropic opened a research preview of what it calls the Model Hardware Standard, or MHS, according to the Financial Times as reported by pharmaphorum and confirmed by NDTV Profit. MHS lets Claude directly operate lab equipment: microscopes, liquid handlers, lasers, robotic arms.

According to Anthropic's Jonah Cool speaking to NDTV Profit, a lot of science simply doesn't happen because the equipment is too technically demanding to operate. MHS gives any AI agent a standardized way to read and write commands to hardware, the same way Anthropic's earlier Model Context Protocol let Claude and rival ChatGPT plug into Gmail or Slack.

Anthropic has been quietly testing this with Amazon Web Services, Danaher, Hugging Face, Raspberry Pi, Genentech, Carnegie Mellon, and the Howard Hughes Medical Institute's Janelia Research Campus, according to NDTV Profit and pharmaphorum. In one promotional demo, a Genentech scientist reportedly sent Claude a PDF of an experiment design and the system autonomously ran it on MHS-equipped hardware.

MHS won't work with hardware lacking a programmable interface, and Anthropic says it wants to harden the system against errors before open-sourcing it, with no firm timeline given, per NDTV Profit.

The Bigger Picture

None of this is happening in a vacuum. pharmaphorum reports the hardware push comes as Anthropic pursues a planned IPO that could value the company at $2 trillion. Crypto Briefing separately cites Anthropic's current estimated valuation at roughly $900 billion. Those are two different numbers describing two different moments, and readers should not confuse Anthropic's present valuation with a future IPO target that has not happened.

Anthropic is also racing OpenAI for the loyalty of academic researchers. nextaipress notes OpenAI launched a comparable "ChatGPT for Academic Researchers" program earlier in the summer of 2026. Google, Nvidia, and others are chasing the same physical-world AI market that a Barclays report cited by NDTV Profit estimates could be worth over a trillion dollars by 2035.

The open questions: which government agency is running the life sciences Mythos access program, how many participants have joined it, and whether Anthropic follows through on its stated intent to expand seats well beyond the initial 10,000. Anthropic has not given a timeline for any of it.