The Pulse
Anthropic mathematician claims Claude helped prove Riemann hypothesis
Levent Alpöge says he and Claude found a proof of the Riemann hypothesis and formalized their argument. The proposed proof has not been independently verified.

Anthropic mathematician Levent Alpöge says he and Claude have found a proof of the Riemann hypothesis, a long-unsolved mathematical conjecture. The proposed proof has not been independently verified, The Verge reports.
In a statement posted on his website, Alpöge said he believed he had discovered the proof working with Claude. He also said the method extends to the Generalized Riemann Hypothesis and expressed an expectation that it could reach the Extended Riemann Hypothesis with more time.
The proposed argument
The arguments were formalized without assumptions, Alpöge said in the statement, an extract of which was published by Simon Willison on November 20. The mathematician credited Claude with outpacing him near the end of the work.
The proposed solution is presented in Lean, a language used to verify mathematical arguments, according to The Verge. The outlet reports that Anthropic had not commented or involved itself in Alpöge’s announcement.
Scientific American reports that specialists will need to check whether the hypothesis has been properly coded and whether the proof is correct. Alpöge’s claim about the generalized hypothesis is also unverified.
The problem and the prize
The conjecture dates to 1859, when mathematician Bernhard Riemann proposed it, Scientific American reports. It concerns the Riemann zeta function and the distribution of prime numbers.
Anthropic’s August research post explains that the zeta function describes the distribution of primes, and that the hypothesis holds that the zeros determining those primes all lie along a particular vertical line. The company said at the time that no one had proved or disproved the hypothesis.
According to The Verge, the problem is one of seven Millennium Prize problems administered by the Clay Mathematics Institute, which offers a $1 million prize for each problem solved. Alpöge’s proposed solution has not been publicly accepted by the institute.
The outlet reports that a proposed solution must be published, withstand at least two years of scrutiny and receive general acceptance in the global mathematics community before it can be considered for the prize.
Claude’s earlier attempt
In August, Anthropic described an earlier attempt by an unreleased research version of Claude to tackle the hypothesis. The company said the model did not solve it, but found a result concerning a related question: the minimum proportion of zeros known to lie on the relevant line.
The company said the model increased that lower bound from 41.6% to 67.2%. Anthropic said it did not expect the techniques used in that earlier work to lead to a proof of the hypothesis.
The earlier result drew on work by Aryan; Baluyot, Goldston, Suriajaya and Turnage-Butterbaugh; and a 2000 paper by Bombieri, according to the company. Anthropic said Claude combined those results to reach the new bound.
How the earlier work was checked
Anthropic said the earlier work took two sessions in Claude Code and used 31 million output tokens. Staff member Jarred Sumner, whom the company described as a non-mathematician, prompted the model to attempt the hypothesis and left the mathematical choices to it.
The company said the model initially tried 650 ideas without success, then spent a day and a half coordinating about 60 Claude subagents. Those agents ran numerical checks and reviewed one another’s work.
Alpöge and fellow Anthropic mathematician Ralph Furman examined the earlier result, the company said. Claude also worked with staff member Eric Easley on a Lean formalization of that result.
Sources
- Original source Anthropic · page 1 alpo.ge
Other pages read
- Anthropic · page 2 anthropic.com
- The Verge theverge.com
- Scientific American scientificamerican.com
- Simon Willison’s Weblog simonwillison.net
Companies in this story
Related stories
- Anthropic says Claude solved two black-hole physics problems
- Anthropic resolves Claude errors after API and Code disruption
- Anthropic advises deleting "think hard" lines from Claude Opus 5.5 prompts
- Anthropic opens self-hosted Claude Managed Agents to private beta
- Anthropic publishes recommendations for designing, running and reporting AI safety tests