Shelah’s eventual categoricity and the prescribed-threshold obstruction
OpenAI groups related papers into a “family”. OpenAI’s catalogue describes this one:
Proves Shelah’s eventual categoricity conjecture in ZFC: for each bound on the Löwenheim–Skolem number, a uniform threshold makes categoricity of an abstract elementary class in one cardinal above that threshold imply categoricity throughout the same tail. Categoricity means uniqueness up to isomorphism at a given cardinality. Under the continuum hypothesis, a proposed specific Hanf threshold need not suffice. (OpenAI’s Lean note)
Lean is a language for mathematical statements and proofs that a computer can verify. The assessments below ask whether those statements express the result OpenAI describes here.
The independent assessments find that the Lean statements express only part of the result described above, or a narrower statement.Read the assessments
Possible match to OpenAI’s description, selected from the papers’ main statements because it shares the most wording. No review confirming the selection is recorded.
The other possible matches are Theorem 1.2(Uniform eventual categoricity) in Eventual categoricity for abstract elementary classes () and Theorem 1.3(The qualitative form) in Eventual categoricity for abstract elementary classes ().
Assume the continuum hypothesis. There is an abstract elementary class K in a countable finitary relational language such that
(i)
\LS(K)=\aleph_0;
(ii)
K has at least two nonisomorphic models of cardinality \beth_{\omega_2}=H(K);
(iii)
putting \Lambda=\beth_{(2^{\aleph_1})^+}, the class K is categorical in every cardinal \mu\ge\Lambda.
In particular, \lambda=\Lambda^+>H(K) is a categoricity cardinal from which downward transfer to H(K) fails.
Reported proof verification
One Lean statement in the release; three independent groups report that Comparator accepted runs for this file at the version shown here.
Assuming the continuum hypothesis, we construct an abstract elementary class with Löwenheim–Skolem number ℵ0 that is categorical in every sufficiently large cardinal but has at least two nonisomorphic models of cardinality \(\beth_{\omega_2}\). Thus categoricity does not transfer down to the proposed bound \(\beth_{(2^{\aleph_0})^+}\), which equals \(\beth_{\omega_2}\) under CH. Consequently, if ZFC is consistent, the prescribed-threshold form of Shelah’s categoricity conjecture is not provable in ZFC.
We prove Shelah’s eventual categoricity conjecture for abstract elementary classes in ZFC. For each infinite bound on the Löwenheim–Skolem number there is a uniform threshold such that categoricity in any one cardinal at or above that threshold implies categoricity in every cardinal at or above the same threshold.
One Lean statement in this family
Lean is a language for mathematical statements and proofs that a computer can verify. The Lean statement below is shown beside its possible paper counterpart, with the reason they were linked.
For a configured theorem, the Comparator program tests whether OpenAI’s separate proof proves the exact Lean statement using only the allowed axioms. Some statement files instead configure definitions for comparison by name and type. Acceptance does not establish that a formal statement means the same thing as the paper’s theorem.
This file states the theorem with sorry, a placeholder, in place of its proof. OpenAI supplies a separate proof file. Comparator tests whether that proof proves the statement here.
three independent groups report that Comparator accepted runs for this Lean statement file: cert-machine; MathVet; Neal McBurnett, openai-math-lean-tooling. Who ran the proof-verification program
cert-machine records it as “CERTIFIED”.
MathVet records it as “comparator-sandboxed-pass”.
Neal McBurnett, openai-math-lean-tooling records it as “accepted”.
Other mathematical statements in OpenAI’s Lean library
Atlas found these propositions by scanning the text of the project’s Model.lean file. Each name links to its definition, with any associated statement file beside it. This list does not record running Lean to interpret or verify those definitions.
Independent comparisons and proof-verification reports
Does the Lean express the advertised result?
Neither group finds that the Lean statements express or imply the whole advertised result. They describe what the statements express in different terms.
MathVet and Carlos Toledo’s cert-machine independently assessed whether this family’s Lean statements express the whole result OpenAI describes in its catalogue. These are assessments of the family as a whole, not of each paper theorem separately.
MathVet finds that the Lean statements express part of the advertised result (their term: “partial”).
Their note:PRIMARY CLAIM NOT STATED (only a secondary result is formalized). CHObstruction.main (hCH : CH) : ∃ L countable relational, ∃ K, K.IsAEC ∧ K.HasLSNumber ℵ₀ ∧ endpoint = hanf ℵ₀ ∧ K.TwoModels endpoint ∧ ∀ μ, tailThreshold ≤ μ → K.Categorical μ is the summary’s last sentence (under CH a prescribed Hanf threshold need not suffice), formalized with hand-written AEC axioms on ZFSet models. The headline, Shelah’s eventual categoricity conjecture proved in ZFC, is not stated (docs: 'The later canonical-point obstruction and the separate eventual-categoricity theorem are not included').
Definitions the group asks readers to compare with the paper:Hand-rolled abstract elementary classes: structure ClassData (objects, strong), IsAEC (iso-closure, partial order, coherence, chain unions over arbitrary γ.ToType, LS bound), LSBound/HasLSNumber, Categorical μ := (∃ M, objects M ∧ card = μ) ∧ ∀ M N of size μ, Nonempty (M.Iso N), hanf κ := beth (succ (2^κ)).ord; standard Shelah axioms but entirely custom, and antisymm is literal equality M = N of RelModels.
“partial”, in their words: Only a special case is stated, or a co-equal headline claim has no challenge statement.
Their method: An independent, model-assisted, pre-referee reading of each family’s challenge statements against its headline claim under a published rubric (PROTOCOL.md), with its own Comparator runs per challenge.
cert-machine finds that the Lean statements express a narrower statement than the advertised result (their term: “NARROWER — DECLARED”).
The gap they name:The headline’s main claim — Shelah’s eventual categoricity conjecture proved in ZFC (uniform threshold, transfer of categoricity through the tail) — is not stated; only the CH obstruction to the prescribed Hanf threshold is.
“NARROWER — DECLARED”, in their words: The challenge statements state less than the headline, and OpenAI’s Lean note says so.
Their method: Every Comparator challenge re-run on Linux with nanoda switched on, decided by a preregistered rule.
Under the continuum hypothesis, the formalized result refutes the proposed transfer of categoricity down to the Hanf threshold. It gives an abstract elementary class in a countable relational language, with Löwenheim–Skolem number $\aleph_0$, that has two nonisomorphic models at $\beth_{\omega_2}=h(\aleph_0)$ but is categorical in every cardinal at least $\beth_{(2^{\aleph_1})^+}$.
No amalgamation, joint embedding, tameness, or absence-of-maximal-models assumption is imposed. The later canonical-point obstruction and the separate eventual-categoricity theorem are not included.
AI-drafted comparisons, with no recorded human review
Unreviewed AI draft: states a paper’s theorem as the paper does. States the main theorem: under CH, an abstract elementary class with two nonisomorphic models at the Hanf threshold that is categorical in all large cardinals.
Who ran the proof-verification program?
These groups report running Comparator with the release’s Lean statement files. For a configured theorem, it verifies OpenAI’s separate proof against the exact statement; some files instead configure definitions for comparison by name and type. Each row is the named group’s report. Atlas has not repeated the runs.
These reports do not list the software versions, proof-verification programs, axioms, time or memory in separate fields. Any such details in a report’s prose are reproduced below.
What these papers cite in the release, and what cites them
These links come from the papers’ bibliographies. A citation records that one paper cites another; how each cited result is used has not been examined here.
Papers elsewhere in the release that this family’s papers cite
This family’s papers cite no paper elsewhere in the release.
Papers elsewhere in the release that cite this family’s papers
No paper elsewhere in the release cites this family’s papers.
References outside the release
Each paper’s outside references are listed with it on its reading page, where each citation is placed. Every outside work cited in the release: References outside the release
Explain a result, compare statements or propose a correction
Atlas has no contributed explanation for this family yet. You can offer one with a GitHub account. You receive named credit and disclose any AI assistance.
Each button opens a GitHub form with this page and its source version filled in. Opening the form does not reserve the work.
Commentary on this family by others, dated and credited. It is commentary, not a finding.
An unsigned public tier list, the OpenAI Math Tier List, classes it as a proof and rates its significance, if it holds up, as “Career-defining publication”. The page says that nothing on it is verified.
VibeMathed, VibeMathed: math problems solved with AI, 11 October 2026. Source
Contested: the release says its construction contradicts Espindola’s published Theorem 4.1 on categoricity transfer.
Michael Harris, Now all mathematicians can be Mechanical Turks, 8 October 2026. Source
Reader comment: paper 240 on Shelah’s eventual categoricity uses non-standard notation, invents words and omits attributions.
Questions awaiting review
Each item identifies a statement, definition or report that needs further review. Its link opens a GitHub form with the source version already recorded.
Read the Lean statement against the paper
An AI agent drafted an assessment of which part of the paper this Lean statement expresses. No named human review is recorded.Open the form
Identify the theorem the catalogue describes
Family 240The theorem shown was selected from shared wording with OpenAI’s description. Atlas records no review confirming that it is the theorem the description means.Open the form
Address a question raised in an independent assessment
Only a special case is stated, or a co-equal headline claim has no challenge statement.
Their method
An independent, model-assisted, pre-referee reading of each family’s challenge statements against its headline claim under a published rubric (PROTOCOL.md), with its own Comparator runs per challenge. Its protocol describes a planned error-rate study; this import does not establish that the study has been completed.
The challenge statements state less than the headline, and OpenAI’s Lean note says so.
Their method
Every Comparator challenge re-run on Linux with nanoda switched on, decided by a preregistered rule; every family’s challenge statements read against OpenAI’s headline, clause by clause. Decider: Comparator (leanprover/comparator d03acab1 = tag v4.34.0, built with the release’s toolchain v4.34.1), lean4export 076e8e57, landrun 811cfff5, nanoda_lib 3a240721 — on GitHub’s Linux runners…
Re-run through Comparator in a sandboxed virtual machine: accepted.
Date
8 October 2026
Version
at the release’s recorded version
Their method
An independent, model-assisted, pre-referee reading of each family’s challenge statements against its headline claim under a published rubric (PROTOCOL.md), with its own Comparator runs per challenge.
Re-run through Comparator with Lean’s kernel: accepted (the check took 51 s; peak memory 6.1 GiB).
Date
10 October 2026
Version
at the release’s recorded version
Their method
Each challenge’s solution is built and checked by Comparator at the pinned release, on Lean leanprover/lean4:v4.34.1, recording the verdict, the check time and the peak memory.
three independent groups report that Comparator accepted runs for this Lean statement file: cert-machine; MathVet; Neal McBurnett, openai-math-lean-tooling.