Alpöge's Riemann Hypothesis Lean proof independently verified in 2026?
This market asks whether someone outside Levent Alpöge's circle and Anthropic publicly shows that his Lean formalization of a Riemann Hypothesis proof passes a defined technical check. The check is a run of the official Comparator tool plus a second proof checker. The market is only about this formal check, not about whether mathematicians accept the proof.
Right now traders give this a 96% chance.
| Market | Chance of yes | Deadline |
|---|---|---|
| Will Alpöge's Lean proof of the Riemann Hypothesis pass independent verification by December 31, 2026? | 96.5% | Jan 1, 2027, 4:59 UTC |
Odds read from Polymarket at 22:18 UTC on Nov 21, 2026. $12K traded so far, $11K in the last 24 hours.
How this market resolves
Yes: Between November 20, 2026 and December 31, 2026, 11:59 PM ET, an independent verifier publishes a report of a successful Comparator run on the formalization exactly as published, using its pinned Mathlib. The run must check the five named theorems against src/Challenge.lean at the stated commit and allow only the axioms propext, Quot.sound and Classical.choice. Lean's kernel (v4.36.0 or later) and at least one other public Lean checker must accept it. The report must include the output and enough detail to reproduce the run.
No: No qualifying report is published by the deadline. This includes the case where only builds, unsupported claims, or checks of a revised version appear, or where the report's output is credibly shown to be fabricated or its hashes do not match its inputs. A reported failed verification does not end the market early, but it also does not make it Yes.
When it really ends
December 31, 2026, 11:59 PM ET. In winter that is UTC-5, so this is 2027-01-01 04:59 UTC. The data's end date of 2027-01-01T04:59:00Z matches the rules.
What counts
- A run of the official 'lake comparator' included in an official Lean release, or of a tagged release of leanprover/comparator.
- A report whose named authors include no one who is Levent Alpöge or an Anthropic employee or contractor at the time of the report.
- Adding the challenge file and the configuration needed to run Comparator, as done in the Deicyde/riemann-hypothesis repository.
- A second checker such as nanoda or leanchecker-paranoid; either one counts.
- A qualifying report even if another checker rejects the formalization, unless the output is shown to be fabricated or altered.
What does not count
- A successful build alone.
- An unsupported claim that the proof was verified.
- A verification of a revised or different version of the formalization.
- A report written by Alpöge or by an Anthropic employee or contractor.
- A report whose output is credibly shown to be fabricated or altered, or whose published hashes do not match its stated inputs.
- Broader mathematical acceptance, peer review, or recognition by the Clay Mathematics Institute.
Fine print people miss
- The formalization must be the archive with sha256 9631b4f0...c7fc, mirrored at commit 40e56bab026552e089d7afcfeff08d2defb9c005 of Deicyde/riemann-hypothesis.
- The statements are checked against src/Challenge.lean at commit 1f2beda145f7b7a9f6f6f6640a60f2e44920b7ba, and riemannHypothesis must be stated as Mathlib's RiemannHypothesis.
- The five theorems checked are riemannHypothesis, riemannZeta_eq_zero_re, LFunction_ne_zero_of_half_lt_re, LFunction_eq_zero_re and LFunction_eq_zero_re_of_isPrimitive.
- Build, export and sandbox steps, other tool versions, and the set of further checkers may differ from the repository's setup.
- A further checker that cannot read the formalization for reasons unrelated to its correctness, such as an unsupported inductive type or export format, does not block a qualifying verification.
- The report needs at least one named author and a public timestamp within the window. The market resolves as soon as a qualifying report appears. A consensus of credible technical reporting may also be used as a source.
What moves the price
- A well-known Lean developer or group publishing a full Comparator run with commands, versions and hashes would push toward Yes.
- A reported Comparator failure, an extra axiom, or a mismatch with the Challenge.lean statements would push toward No.
- Evidence that the challenge statements do not match Mathlib's RiemannHypothesis would push toward No.
- Reports of a second checker, such as nanoda or leanchecker-paranoid, accepting the proof would push toward Yes.
- Evidence that a published report was fabricated or that its hashes do not match would push toward No.
- Fixes or new versions of the formalization would push toward No, since only the exact published version qualifies.
Put a bot on this market
A PROWL bot reads the news around the clock, judges each post against these exact rules, and trades inside the limits you set. It starts in paper mode.
Questions people ask
What does it take for this Riemann Hypothesis Lean market to resolve Yes?
An independent party must publish a report showing a successful run of the official Comparator on Alpöge's formalization exactly as published. A second Lean checker must also accept it. The report must be public by December 31, 2026, 11:59 PM ET.
Does this market decide whether the Riemann Hypothesis is proven?
No. The rules say it covers only whether this specific formalization passes the defined independent technical verification. It does not cover peer review, wider mathematical acceptance, or the Clay Mathematics Institute.
Who counts as an independent verifier?
A verifier is independent if none of the report's named authors is Levent Alpöge or, at the time of the report, an employee or contractor of Anthropic. The report must name at least one author.
Does a successful Lean build count as verification?
No. A build alone does not qualify. The report needs a Comparator run that checks the five named theorems, allows only three axioms, and is accepted by the Lean kernel and another checker.
What happens if a verifier reports that the check failed?
A reported failure does not by itself resolve the market before December 31, 2026. The market stays open, and a later qualifying report can still make it Yes. A rejection from another checker does not cancel a qualifying report, unless the output is shown to be fabricated or altered.
When does this market end?
It ends on December 31, 2026 at 11:59 PM ET, which is 2027-01-01 04:59 UTC. It can resolve earlier if a qualifying report is published before then.
More Science markets explained
Written from the market's own rules text on Nov 21, 2026. If the rules on Polymarket differ from this page, the rules on Polymarket govern.