A 167-Year-Old Maths Mystery May Have Been Solved. An Anthropic Mathematician Says Claude Helped

Last Updated:
Levent Alpöge says he and Claude found a proof of the Riemann hypothesis, a puzzle about how prime numbers are scattered along the number line. Two independent computer checkers have accepted the proposed proof. Mathematicians are still trying to understand why it works
When Claude tried to explain its proposed proof of the Riemann hypothesis in plain language, the draft ran to more than 250,000 pages. Mathematician Levent Alpöge published the code instead
When Claude tried to explain its proposed proof of the Riemann hypothesis in plain language, the draft ran to more than 250,000 pages. Mathematician Levent Alpöge published the code instead Credits: Getty

Levent Alpöge needed four pages to tell the world what he believed he had done. Explaining how he had done it was another matter. An attempt to get Claude to put the mathematics into prose ran to more than 250,000 pages. The Anthropic mathematician decided to publish the code instead. That ran to roughly 17 million lines.

Sign up for Open Magazine's ad-free experience
Enjoy uninterrupted access to premium content and insights.

“I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis,” Alpöge wrote in a note posted on his personal website on November 20. He made a bigger claim in the next sentence. The same method, he said, also establishes the generalised Riemann hypothesis, which extends the question to a wider family of mathematical functions.

open magazine cover
Open Magazine Latest Edition is Out Now!

The Wealth Issue 2026

06 Nov 2026 - Vol 05 | Issue 45

Portraits of Dreamers and Doers under 35

Read Now The Wealth Issue 2026

Within hours, MIT mathematician Andrew Sutherland had compiled all 175,187 modules of Alpöge’s submission without an error. By 3.35 am IST on November 21, his public log recorded a stronger result. Two independently written proof checkers had accepted the full exported argument. Further checks were still running. The Clay Mathematics Institute, which offers $1 million for a solution, continued to list the problem as unsolved.

What Riemann saw in the primes

Start with 11 and count the gaps between prime numbers. From 11 to 13, the gap is two. From 13 to 17, it is four. Then 17 and 19 sit close together again. Farther along the number line, primes grow scarcer on average, though close pairs still turn up. Mathematicians have a reliable estimate for how many primes to expect below a large number. The harder question is how far the actual count can stray from that estimate.

In 1859, German mathematician Bernhard Riemann connected that question to a mathematical function that now bears his name. The connection is surprising. To understand how primes are scattered along a line, you can look at a picture in two dimensions. Plot the points where Riemann’s zeta function equals zero. His hypothesis says that every zero mathematicians call non-trivial sits on a single vertical line, one-half unit to the right of the origin. If that holds, it puts a tight limit on how far the prime count can stray.

Mathematician Levent Alpöge says Claude helped him find a proof of the Riemann hypothesis, a 167-year-old prediction about how closely the count of prime numbers follows a pattern
Mathematician Levent Alpöge says Claude helped him find a proof of the Riemann hypothesis, a 167-year-old prediction about how closely the count of prime numbers follows a pattern Credits: Getty Images

The first 10 trillion non-trivial zeros have been checked. Every one fell on the line. That still leaves room for a zero somewhere farther out to sit a fraction to one side. One such exception would bring down the hypothesis. A proof has to establish why there can never be one.

The moment Claude pulled ahead

As a student, Alpöge imagined his own attempt on Riemann lay decades ahead. In the meantime, he set about learning any mathematical tool he thought might come in handy. When he eventually worked on the problem with Claude, they began with a simpler version in which two seemingly different ideas could be brought together. Alpöge says he recognised where that combination might work in the harder problem and steered Claude towards it. Then, by his account, the balance of the partnership shifted.

“Claude flew, and danced, and sparkled near that finish line,” he wrote.

There had been an early sign of progress in August. Anthropic reported that a research version of Claude had raised the minimum proportion of zeros that mathematicians could prove lie on Riemann’s line. Two mathematicians at the company examined the argument and wrote a shorter account for specialists. What separates that result from November’s claim is a single word. Alpöge now says he can account for all the non-trivial zeros.

What the computers have checked

Alpöge released his argument in Lean, a programming language that lets mathematicians write a theorem in precise terms and check each step used to reach it. That gave other researchers a way to test his work the day it appeared. Sutherland at MIT and Simon DeDeo at Carnegie Mellon University independently compiled the code, Quanta Magazine reported.

One question in an exercise of this size sounds almost too simple to ask. Does the statement proved by the code say exactly what mathematicians mean? Change a definition, even slightly, and a perfectly logical argument could establish something easier. Sutherland’s log says an independent comparison matched all five headline theorems to their intended statements. At the time of his update, the official comparison procedure and a separate replay of the exported proof were still running.

Even if every check passes, a question remains for anyone hoping to use the discovery. Which idea makes the argument work? A mathematician who grasps that idea might simplify the proof, teach it or recognise where else it applies. New York University number theorist Maksym Radziwiłł put his reservation plainly to Quanta. “The object of mathematics as the rest of science is really the question ‘why?’ and this doesn’t answer it at all.”

A second proposed proof, in 940 pages

While researchers were checking Alpöge’s code, Ravi Vakil, president of the American Mathematical Society, released a 940-page manuscript by Stanford postdoctoral scholar Alessandro Masullo. It proposed a separate proof of a generalised form of the Riemann hypothesis. According to Vakil, Masullo had worked without AI. Stanford mathematician Brian Conrad had already spent months reading the argument and giving him feedback.

The plan had been to give Masullo time to polish the manuscript and circulate it among specialists. Alpöge’s announcement changed Vakil’s mind about waiting. He released the draft so other mathematicians could examine it. It remains a proposed proof.

The million dollars can wait. Under Clay’s rules, a solution must appear in a qualifying publication and gain general acceptance among mathematicians worldwide. At least two years must pass after publication before the institute will consider it for the prize.

For now, the two proposals have given mathematicians an unusual reading assignment. One runs to 940 pages and has already occupied a colleague for months. The other can be checked by computers in hours, yet its author is still trying to find a shorter way to explain it. After 167 years, that may be the most intriguing prospect of all: an argument the next student who falls for Riemann’s puzzle can read, understand and use.