Key takeaways
- The Riemann hypothesis says the nontrivial zeros of the zeta function have real part 1/2; the generalized version says the same for every Dirichlet L-function.
- The zeta function is the Dirichlet L-function of the trivial character, so GRH implies RH, while RH alone does not give GRH.
- On November 20, 2026, Levent Alpöge claimed a Lean-only proof of RH and GRH made with Claude; as of November 21 nobody has verified it.
Riemann hypothesis vs generalized Riemann hypothesis comes down to scope. The Riemann hypothesis (RH) is about the zeros of one function, the Riemann zeta function. The generalized Riemann hypothesis (GRH) makes the same claim for an infinite family, the Dirichlet L-functions, of which zeta is the simplest member. So GRH implies RH, but a proof of RH alone would not give GRH.
This matters now because on November 20, 2026, Levent Alpöge, a mathematician at Anthropic, wrote that he and Claude found a proof of RH and that the method also settles GRH. Nobody has verified that claim yet.
What the Riemann hypothesis says
The zeta function starts as the sum 1 + 1/2^s + 1/3^s and so on, extended by Riemann to almost every complex number s. It is zero at -2, -4, -6 and the other negative even numbers, the trivial zeros. The official Clay problem description by E. Bombieri states the hypothesis in one line: every nontrivial zero has real part 1/2, so it sits on a vertical line called the critical line. Those zeros control how far the count of primes strays from its average. Our explainer on what the Riemann hypothesis is covers that link and the $1 million prize.
What the generalized Riemann hypothesis adds
Give each term of the zeta sum a weight that repeats in a fixed pattern. For example: +1 for numbers that leave remainder 1 when divided by 4, -1 for remainder 3, and 0 for even numbers. That weighting is a Dirichlet character. Wikipedia's GRH article gives the rules: the pattern repeats every q numbers, the weight of a product is the product of the weights, and the weight is zero exactly on numbers sharing a factor with q.
Each character χ gives its own Dirichlet L-function, the sum of χ(n)/n^s. There is one for every pattern and every q, so infinitely many. Dirichlet used this idea in 1837 to prove that progressions like 3, 7, 11, 15, 19 contain infinitely many primes, and these functions are still how mathematicians count primes in such progressions.
GRH says every nontrivial zero of every such function, for every primitive character, has real part 1/2. A primitive character is one not copied from a pattern with a shorter period; the condition rules out extra zeros that are off the line for trivial reasons. Wikipedia, citing Davenport's textbook Multiplicative Number Theory, says Adolf Piltz probably first stated GRH in 1884.
Why GRH implies RH, but not the reverse
Set every weight to 1. The trivial character χ(n) = 1 turns the L-function back into zeta, and Wikipedia notes that this case yields the ordinary Riemann hypothesis. Anyone who proves GRH has proved RH. Knowing where zeta's zeros are, though, says nothing direct about the other L-functions.
The two may not differ much in difficulty. Bombieri writes that RH for zeta does not seem any easier than for Dirichlet L-functions, except possibly for one special kind of real zero, which suggests a proof may need to attack the general problem. Alpöge's claim fits that: he presents GRH as falling to the same method.
RH, GRH, ERH: the names
Coverage mixes these names up. Wikipedia lays out the usual meaning:
| Name | Which functions | Relation |
|---|---|---|
| Riemann hypothesis (RH) | Riemann zeta function | The Clay Millennium Prize problem |
| Generalized (GRH) | Dirichlet L-functions | Includes RH |
| Extended (ERH) | Dedekind zeta functions of number fields | Implies GRH |
| "GRH" in the broad sense | All global L-functions | The widest version |
This matters for reading Alpöge's statement. He says the method resolves "the so-called Generalized Riemann Hypothesis" and that he expects to reach the Extended Riemann Hypothesis with more time. He doesn't define the terms, but by the usual naming the Dedekind zeta case is not yet claimed.
What depends on GRH
Many theorems are written "assuming GRH." Some the sources list:
- Primes in progressions. Under GRH, primes split evenly among progressions with an error of about the square root of x (Wikipedia). RH gives this only for all primes together.
- Miller's primality test. G. Miller showed in 1976 that, assuming a form of the hypothesis, checking a number n against every base up to 2(ln n)² decides whether it is prime in polynomial time, according to Wikipedia. Bombieri lists the test among the consequences.
- Weak Goldbach. Hardy and Littlewood showed in 1923 that GRH implies every large enough odd number is a sum of three primes; a 1997 paper by Deshouillers, Effinger, te Riele and Zinoviev showed that under GRH this holds for every odd number above 5.
- Artin's primitive root conjecture, which Hooley showed in 1967 follows from GRH.
- Number fields. Bombieri lists a very strong form of Chebotarev's density theorem, about how primes split in number fields.
Terence Tao noted in a comment on his November 20 post that the Landau-Siegel problem also follows from GRH.
Some of these were later proved without it. The AKS test of 2002 checks primality in polynomial time unconditionally, and Vinogradov proved weak Goldbach for large odd numbers in 1937. Bombieri counts such cases as evidence the hypothesis is true.
What Alpöge claims
His four-page statement says he believes he found a proof of RH with Claude, that it extends to GRH, and that the two formalized it in Lean, a language where a computer checks each step, with no assumptions. He released only the Lean code, in 14 archive files, and promises a proper write-up. A timeline he says Claude compiled ends "11/16/26: GRH." Scott Aaronson put the code at 17.4 million lines.
The methods he names, such as theta lifts and automorphic forms on higher-rank groups, come from the broad L-function world. Tao wrote that the proof, from very initial analysis, "appears to use the full power of modern developments on the Langlands program and arithmetic geometry." Our news story on the claim covers the announcement.
How mathematicians have reacted
As of November 21, no source says the proof is verified.
- Scientific American reported that MIT's Andrew Sutherland compiled the Lean code independently and called it a historic moment, while saying the mathematics still has to be understood. Experts must also check that the Lean statement really encodes RH and GRH.
- Tao called the GRH proof the least digested of three recent AI-linked results and pointed to the Lean community's Zulip chat for verification work.
- Aaronson wrote that he isn't 100% sure but bets the proof will stand, while noting Lean has had kernel bugs.
The same day, Ravi Vakil posted a separate GRH draft by Alessandro Masullo, written without AI and reviewed for months by Brian Conrad. It is unverified too. Claude's earlier, checked result was smaller: Anthropic reported in August that it raised the proven share of zeta zeros on the critical line from 41.6% to 67.2%.
What would count as solved
The Clay Institute's prize rules require publication in a qualifying outlet, at least two years after publication, and general acceptance by the mathematics community. The prize problem is RH, which GRH contains, so a GRH proof that passes those tests would cover it. The next things to watch are the Lean checks and Alpöge's promised paper.
Frequently asked questions
Is the Riemann hypothesis solved?
Not as of November 21, 2026. Levent Alpöge claimed on November 20 that he and Claude proved RH and GRH, posting only Lean code, and Ravi Vakil posted a separate human-written GRH draft by Alessandro Masullo. Neither has been verified.
Does the generalized Riemann hypothesis imply the Riemann hypothesis?
Yes. The zeta function is the Dirichlet L-function of the trivial character χ(n) = 1, so RH is one case of GRH. The reverse does not hold.
What does the Riemann hypothesis have to do with cryptography?
The best-known link is primality testing: Miller's 1976 test is deterministic and fast if a form of the generalized hypothesis holds. In practice the probabilistic Miller-Rabin test is used, and the AKS test has run in polynomial time without any hypothesis since 2002. None of the sources say a proof would affect encryption in use today.
Is the generalized Riemann hypothesis a Millennium Prize problem?
The Clay Millennium Prize problem is the Riemann hypothesis for the zeta function. GRH is not a separate prize, but since it includes RH, a GRH proof that meets Clay's rules of publication, two years and general acceptance would cover the prize problem.
Sources
- Remark (statement on the Riemann Hypothesis)Levent Alpöge
- Problems of the Millennium: the Riemann Hypothesis (official problem description)Clay Mathematics Institute
- Rules for the Millennium PrizesClay Mathematics Institute
- Generalized Riemann hypothesisWikipedia
- Miller-Rabin primality testWikipedia
- Unforced Euler singularity, unforced Navier-Stokes singularity, and the generalized Riemann hypothesisTerence Tao
- Hurried announcement: Alessandro Masullo GRHRavi Vakil
- The stochastic parrots have apparently proved the Riemann HypothesisScott Aaronson, Shtetl-Optimized
- Anthropic mathematician claims to prove Riemann hypothesis, most famous problem in all of mathScientific American
- Claude has improved on a longstanding lower bound for the fraction of zeros of the Riemann zeta functionAnthropic


