Skip to content
FR · Français
Get started
Registry

AlphaProof Nexus (Google DeepMind, Science): 9 of 353 open Erdős problems and 44 of 492 OEIS conjectures solved in Lean, according to the authors

Status

Peer reviewed

Recorded on · api.crossref.org

  1. Announced
  2. Revisednot evaluated
  3. Revisedpeer reviewed
What this status means

Source reading : Source read word for word (level 3 of 3)

Published in a journal or reviewed by named experts.

Full history

The gist

Google DeepMind published its AlphaProof Nexus study in Science on 8 October 2026: the authors report 9 of 353 open Erdős problems and 44 of 492 OEIS conjectures solved in Lean.

FollowReport

Cite this entry

The status has changed 2 times since Oct 11, 2026: a citation always points to a dated snapshot.

Cite snapshot 3, Oct 13, 2026

APA citation with the dated link

DiscoverIA. (2026, October 13,). AlphaProof Nexus (Google DeepMind, Science): 9 of 353 open Erdős problems and 44 of 492 OEIS conjectures solved in Lean, according to the authors [Status : Peer reviewed, snapshot 3]. DiscoverIA registry. https://discoveria.ai/en/fiche/alphaproof-nexus-science-2026-10-08/#releve-3

BibTeX @misc, key discoveria-alphaproof-nexus-science-2026-10-08-r3

@misc{discoveria-alphaproof-nexus-science-2026-10-08-r3,
  author       = {{DiscoverIA}},
  title        = {AlphaProof Nexus (Google DeepMind, Science): 9 of 353 open Erdős problems and 44 of 492 OEIS conjectures solved in Lean, according to the authors},
  howpublished = {DiscoverIA registry, entry alphaproof-nexus-science-2026-10-08},
  note         = {Status « Peer reviewed », snapshot 3, recorded on October 13, 2026. new result},
  year         = {2026},
  url          = {https://discoveria.ai/en/fiche/alphaproof-nexus-science-2026-10-08/#releve-3},
  urldate      = {2026-10-13}
}

Dated link opens this snapshot in the history

https://discoveria.ai/en/fiche/alphaproof-nexus-science-2026-10-08/#releve-3

Text written by an AI (DiscoverIA pipeline), on Reviewed by: Antoine Olivier, Oct 18, 2026Report an error

What has not been reviewed?

The status and its sources come from the registry; the explanatory text has been reviewed by a human.

The status and its sources come from the registry; the explanatory text has been reviewed by a human.

Key facts

Claimed by
Google DeepMind (AlphaProof Nexus ; 21 auteurs, dont un de l'université d'Aarhus)
announced on Oct 8, 2026
Nature
new result
Field
Mathematics (mathematics / formal proof)
Checked by
Science peer review (anonymous reviewers)
Sources (domains)
7 : 4 read word for word, 2 through an AI summary, 1 links only
This entry
Reviewed by Antoine Olivier, Oct 18, 2026

To understand

Analogy

It is like a record approved by a panel that read the file, but that nobody else has measured again yet.

What it changes

A peer-reviewed article states that an AI system solved mathematical problems that were still open, with proofs written in Lean, a language in which a computer checks every step of a proof. The study first appeared as an arXiv preprint on 21 May 2026.

What remains unclear

The Science reviewers are anonymous and no named third party has rerun the published Lean proofs yet. The figures are the authors' own. For comparison, the OEIS Open benchmark (arXiv 2608.11941) reports 147 of 492 conjectures solved by general-purpose language models.

For experts

7 sources.

For experts

Proofs
article Science (résumé et métadonnées lus via Crossref), préprint arXiv 2605.22763 (texte intégral v2 lu), Lean (dépôt google-deepmind/alphaproof-nexus-results, non rejoué)

Sources(domains)

  • primairedoi.orglink only, reading not recorded
  • source-statutapi.crossref.orgpage read word for word · consulted on Oct 13, 2026 · Advancing mathematics research with AI-driven formal proof search
  • articlearxiv.orgpage read word for word · consulted on Oct 18, 2026 · Our most capable agent autonomously resolved 9 of 353 open Erdős problems
  • articlearxiv.orgpage read word for word · consulted on Oct 13, 2026 · we developed a framework, AlphaProof Nexus, for LLM-aided proof generation
  • leangithub.compage read word for word · consulted on Oct 13, 2026 · Lean math proofs generated by AlphaProof Nexus and accompanying natural language prose proofs.
  • position-revendicateurdigg.compage read through an AI summary · consulted on Oct 11, 2026
  • articlequantamagazine.orgpage read through an AI summary · consulted on Oct 11, 2026

Method

The arXiv preprint 2605.22763 (version 1 of 21 May 2026, version 2 of 8 June 2026) and the Science article (DOI 10.1126/science.aej2213) share the same title and the same 21 authors. The Lean proofs and their natural-language write-ups are published in the google-deepmind/alphaproof-nexus-results repository. Status “peer reviewed” recorded on 13 October 2026 from api.crossref.org (page read, quotation checked word for word). History: 3 dated observations, oldest first.

Limits

A Lean proof supplied by the claimant does not count as formal verification by a third party.

To understand

It is like a record approved by a panel that read the file, but that nobody else has measured again yet.

Status history

The numbers link the band (order of observations) and the thread (most recent first). Band segments have equal widths, not a time scale.

  1. 03

    Not evaluatedbecomesPeer reviewedCurrent status

    peer-reviewed journal article: received 2026-05-27, accepted 2026-08-17, published 2026-10-08 (Science 394(6820):234-239); anonymous reviewers; public Lean proofs not re-run by a…

    Read the full reason

    peer-reviewed journal article: received 2026-05-27, accepted 2026-08-17, published 2026-10-08 (Science 394(6820):234-239); anonymous reviewers; public Lean proofs not re-run by a named third party

    Source: api.crossref.orgread word for word

  2. 02

    Status confirmed:Not evaluated

    observation of the arXiv preprint 2605.22763 (v1 of 2026-05-21)

    Source: quantamagazine.orgread through an AI summary

  3. 01

    First observation:Not evaluated

    Source: digg.comread through an AI summary

An observation is added; none is ever erased.

Report an error or suggest a source

How DiscoverIA uses your address and how to exercise your rights: privacy policy.

Bot check not configured.