Quasi-Riemann Hypothesis Explained: OpenAI’s Proof
Quick answer: The quasi-Riemann hypothesis says every nontrivial zero of the Riemann zeta function stays a fixed positive distance from the edges of the critical strip, where the real part runs from 0 to 1. The same margin must work for every zero. In October 2026, OpenAI released a proof confining the zeros to real parts from 1/8 to 7/8, plus a shorter argument for the wider strip from 1/12 to 11/12. The full Riemann hypothesis puts every zero exactly on the line at 1/2.
Related update, November 21, 2026: the Lean and Mathlib communities have certified Levent Alpöge’s formal proof of the full Riemann Hypothesis. That is separate from OpenAI’s October quasi-Riemann result. This guide explains the weaker statement and what the October work established; for the full story see our Riemann Hypothesis explainer.
In this problem, moving a boundary from 1 to 7/8 is a major change, because the statement must hold for infinitely many zeros with no exceptions, however high up the zeros go. A picture of a road makes the difference between a fixed strip and the center line easy to see.
Background: The Zeta Function and Its Zeros
Write a complex number as \(s = a + bi\). Its real part \(a\) gives its horizontal position in the complex plane. For \(a > 1\), the Riemann zeta function is the convergent series
\[ \zeta(s) = 1 + \frac{1}{2^s} + \frac{1}{3^s} + \cdots \]
Mathematicians extend the function beyond this region by analytic continuation; it has a single pole at \(s = 1\). Its trivial zeros are at \(-2, -4, -6, \ldots\). All other zeros, the nontrivial ones, lie in the critical strip \(0 < a < 1\). The Riemann hypothesis says every nontrivial zero has \(a = \tfrac12\).
What Is the Quasi-Riemann Hypothesis?
Imagine an infinitely long road whose side edges are marked 0 and 1. Each nontrivial zero is a marker somewhere on that road. The quasi-Riemann hypothesis says all the markers fit inside one narrower central band, and the safety margin at each edge stays the same along the entire road.
More precisely, there is one fixed number \(\delta\) with \(0 < \delta < \tfrac12\) such that every nontrivial zero \(\rho\) satisfies
\[ \delta \le \operatorname{Re}(\rho) \le 1 – \delta. \]
The Riemann hypothesis is the much stronger statement that every marker sits on the center line, \(\operatorname{Re}(\rho) = \tfrac12\).
| Statement | Where the nontrivial zeros must lie | Status |
|---|---|---|
| Classical critical strip | \(0 < \operatorname{Re}(\rho) < 1\), with no fixed margin from the edges | Known since the 1890s |
| OpenAI, main result (Sept 30, 2026 manuscript) | \(\tfrac18 \le \operatorname{Re}(\rho) \le \tfrac78\) | Lean formalization of the main zero-free result |
| OpenAI, shorter argument (Oct 5, 2026 manuscript) | \(\tfrac1{12} \le \operatorname{Re}(\rho) \le \tfrac{11}{12}\) | Weaker bound, simpler proof |
| Riemann hypothesis | \(\operatorname{Re}(\rho) = \tfrac12\) | Lean proof certified Nov 21, 2026 |
The 1896 proofs of the Prime Number Theorem showed there are no zeros on the line \(\operatorname{Re}(s) = 1\). Later “zero-free regions” pushed zeros slightly away from that edge, but the guaranteed gap shrinks toward zero as you go up the strip. A fixed margin had never been proved.
Why the 7/8 Bound Is Stronger Than 11/12
A narrower allowed band is a stronger statement. Since \(7/8 = 0.875\) and \(11/12 \approx 0.9167\), the 7/8 result leaves less room for zeros. For example, the 11/12 bound alone would not rule out a hypothetical zero with real part 0.90, but the 7/8 bound would. (This compares what the two bounds establish; it does not describe an actual zero.) So the longer September 30 manuscript proves more, and the shorter October 5 argument is a weaker but easier-to-read alternative.
Why a Fixed Margin Matters for Prime Numbers
Let \(\pi(x)\) count the primes up to \(x\), and let \(\operatorname{Li}(x)\) be the standard estimate of that count. The October 5 manuscript states this consequence of its result:
\[ |\pi(x) – \operatorname{Li}(x)| = O\!\left(x^{11/12} \log x\right). \]
That is an upper bound on how the error can grow, allowing a constant factor, not the exact error at any particular \(x\). The important feature is a fixed exponent below 1, a global prime-counting error estimate of a kind never proved before. The full Riemann hypothesis gives the sharper bound \(O(\sqrt{x}\log x)\), so the two results do not give the same precision.
Mathematician Maksym Radziwiłł told Quanta Magazine that “the vast majority of consequences of RH already follow from quasi-RH.” That is an expert assessment of significance; it does not mean the two statements have identical consequences.
What Exactly Did OpenAI Release?
- October 6, 2026: OpenAI published a large batch of math manuscripts written by an unreleased internal model.
- The main result confines nontrivial zeros to real parts from 1/8 to 7/8. The main zero-free result has a Lean formalization; OpenAI’s documentation notes that the paper’s later applications are not included in that formalization.
- A shorter argument gives the wider 1/12 to 11/12 strip. The formalization label of the main result should not be transferred to it automatically.
- Scope beyond zeta: according to the documentation, the bounds also cover Dirichlet L-functions and certain finite-order Hecke L-functions over \(\mathbb{Q}(\sqrt{-3})\), not every L-function.
- Landau-Siegel zeros: the work also addresses so-called exceptional real zeros extremely close to 1, a long-standing obstacle in number theory. It does not mean “there are no real zeros anywhere.”
How Did Mathematicians React?
In October 2026, Quanta Magazine reported that experts largely agreed the result was correct, while Peter Sarnak said he would not call it a theorem until humans had checked the details. Those comments describe the response at that stage. Six weeks later, the full Riemann hypothesis itself received a certified Lean proof from a different team.
Common Mix-Ups
“AI solved the Riemann Hypothesis in October.” No. October’s result was the quasi version. The full hypothesis was proved in Lean by Levent Alpöge with Claude in November 2026.
“Quasi means approximately true.” No. The quasi-RH is a precise statement with exact bounds; it is simply weaker.
“At least 1/12 away from the edges means on the line.” No. The quasi-RH allows zeros anywhere in the middle band. Only the Riemann Hypothesis pins them to 1/2.
Watch: The Quasi-Riemann Hypothesis Explained in Video
If you learn better by watching, these videos are worth your time.
Check Your Understanding
Five quick questions to see how much of this has stuck. Pick an answer and press Submit; you will see a short explanation either way.
Answer each question and press Submit to see your result right away.
Frequently Asked Questions
Is the quasi-Riemann hypothesis the same as the Riemann hypothesis?
No. The Riemann hypothesis says every nontrivial zero of the zeta function has real part exactly 1/2. The quasi-Riemann hypothesis only says the zeros stay one fixed distance away from the edges of the strip from 0 to 1. It is a weaker statement.
What did OpenAI prove about the quasi-Riemann hypothesis?
In October 2026, OpenAI released work by an internal AI model proving that all nontrivial zeta zeros have real parts between 1/8 and 7/8, with a Lean formalization of that main result, plus a shorter argument for the weaker strip from 1/12 to 11/12.
Has the full Riemann hypothesis been proved too?
Yes, separately. On November 20, 2026, Levent Alpöge released a Lean proof of the full Riemann hypothesis developed with Claude, and on November 21 the Lean and Mathlib communities certified both its statement and its proof.
Why does the quasi-Riemann hypothesis matter?
It gives a prime-counting error bound with a fixed exponent below 1, which had never been proved, and many results previously known only under the Riemann hypothesis follow from it.
Keep Exploring
Sources and Further Reading
Related to This Article
More math articles
- LSAT Necessary vs. Sufficient Assumptions: Identify the Task (2027 Prep)
- Ratio, Proportion and Percentages Puzzle – Challenge 27
- The Best Grade 3 ELA Practice Tests for South Carolina Students
- CD Calculator
- TASC Math Test-Taking Strategies
- Math Exit Ticket Generator
- How to Solve Multi-step Addition Word Problems
- ATI TEAS 7 Math Practice Test Questions
- Dark money
- Countdown Calculator


What people say about "Quasi-Riemann Hypothesis Explained: OpenAI’s Proof - Effortless Math"?
No one replied yet.