Skip to main content
Live
Main content

Anthropic mathematician says Claude helped him prove the Riemann hypothesis

Levent Alpöge released a Lean formalization in 14 archive parts instead of a paper, saying a readable version would run past 250,000 pages.

Jaeden Schafer
Editor in Chief · · 5 min read
Anthropic logo

Levent Alpöge, a mathematician at Anthropic, said Friday that he and Claude have found a proof of the Riemann hypothesis, the 1859 conjecture about how prime numbers are distributed and the most famous open problem in mathematics. He posted the claim in a short statement on his personal website and released the supporting work as computer-checkable code rather than a conventional paper.

Alpöge, who is also a junior fellow at Harvard's Society of Fellows, wrote that the method extends further and resolves the Generalized Riemann Hypothesis, which carries the conjecture over to Dirichlet L-functions. He said he expects, with more time, to address the Extended Riemann Hypothesis as well, but he did not claim that result.

“I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis.”
— Levent Alpöge, Anthropic mathematician

The proof was formalized in Lean, the proof assistant mathematicians use to have software check each logical step. Alpöge released it as 14 compressed archive parts, a package reported at roughly 175,187 source files and 239 MB. He said he and Claude formalized the arguments without assumptions.

Key facts

  • 01Levent Alpöge said on November 20, 2026 that he and Claude found a proof of the Riemann hypothesis, first posed in 1859.
  • 02He says the method extends to the Generalized Riemann Hypothesis; the Extended Riemann Hypothesis is not claimed.
  • 03The proof was released as a Lean formalization in 14 compressed archive parts, with no conventional paper.
  • 04Alpöge estimates a written version of the argument would exceed 250,000 pages.
  • 05The Clay Millennium Prize of $1 million requires peer-reviewed publication and two years of scrutiny before an award.

Alpöge chose not to publish a human-readable write-up. In a footnote, he estimated that a written version of the argument would run past 250,000 pages, and he described a readable version as massively unreadable. His statement itself is about four pages long.

The approach, as Alpöge describes it, builds on Stepanov's method from the function-field setting. Reproducing that method's multiplicity effect over the integers was the central obstacle. The argument relies on repeated theta lifts in an automorphic framework, and Alpöge credits Claude with proposing the length of the chain of lifts.

Alpöge wrote that Claude far outpaced him near the finish line and supplied expertise in automorphic forms that he lacked, while he guided the overall strategy. A timeline in the statement, which he says Claude compiled, dates the key Stepanov determinant step to October 6, 2026, the iterated theta lifts to November 15, and the generalized result to November 16.

“The Riemann Hypothesis is the classic problem in number theory.”
— Levent Alpöge, Anthropic mathematician

The claim follows earlier Anthropic research on the same problem. On August 10, 2026, Anthropic reported that an unreleased research version of Claude raised the proven lower bound on the share of zeta zeros satisfying the hypothesis from 41.6% to 67.2%. That run used two Claude Code sessions, about 31 million output tokens and roughly 60 coordinating subagents, and Anthropic said explicitly at the time that it did not prove the hypothesis.

That August result was examined by Alpöge and Ralph Furman, and the paper was reviewed by number theorists Brian Conrey and Dan Goldston. No comparable outside review of the new proof has been announced. Anthropic has not issued a statement on the claim, and Alpöge's statement thanks mathematicians including Manjul Bhargava and Peter Sarnak without naming them as reviewers.

Related · from this week
Anthropic's unreleased model advances the Riemann hypothesis after 31M tokens
Jaeden Schafer · 5 min read →

Alpöge flagged his own caveats. He wrote that the work has not been checked and that he may have baked lots of strange, suboptimal choices into the argument. The decisive test is whether other mathematicians can confirm that the Lean code compiles in full and that its definitions state the conventional hypothesis.

The Clay Mathematics Institute lists the Riemann hypothesis among its Millennium Prize Problems, with a $1 million award. Its rules require a solution to appear in a qualifying peer-reviewed outlet and survive at least two years of scrutiny, so the claim is not yet eligible.

ShareXLinkedInEmail
AI Box

Every AI model. One chat.

The latest models from ChatGPT, Claude, Gemini, Sora, ElevenLabs — 80+ models in a single chat. Compare answers side by side. Pick the best one every time.

  • ChatGPT, Claude, Gemini, Grok, DeepSeek — in one chat
  • Generate images & video with Sora, Veo, Ideogram
  • Compare any two models side by side
  • From $8.99/mo · 80+ models, all included
Try AI Boxaibox.ai
Trusted by 3,000+ teams
Got a tip?

Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.

Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.

AI Box Daily briefingFree · Daily · No fluff

Stay ahead of everyone in AI.

The tightly edited AI news email engineers, founders, and investors actually open. One email. Every weekday. Five minutes to finish.

Loved by 10,000+ AI professionals
Free forever. Unsubscribe with one click.

The briefing read inside teams at

Keep reading

More from Models

Anthropic logo
Models

Anthropic's unreleased model advances the Riemann hypothesis after 31M tokens

Coordinating 60 subagents across a day and a half, the model tested 650 ideas and improved the known bound on a 150-year-old problem.

Jaeden Schafer5 min read
Anthropic logo
Models

Anthropic details how Claude agents dream to clean up their memory

In Anthropic's CL-Bench test, a memory store lifted Claude Opus 5.5 from 0.655 to 0.838, and idle-time review helped fix bad notes.

Jaeden Schafer5 min read
Anthropic logo
Models

Anthropic launches Claude Docs and Slides to challenge Google's Gemini

Claude now generates shareable documents and presentations from any chat, closing the productivity-suite gap with Google Workspace.

Jaeden Schafer4 min read