explainx.ai350K+ learn from us
TrendingAI News TodayPathwaysSkills
Pricing
explainx.ai

Upskill in AI — 16 free pathways, live workshops & bootcamps, and 50+ courses from practitioners. Plus the skills, tools, and MCP servers to practice on.

follow us

follow on google

Add explainx.ai as a preferred source

corporate training

support@explainx.ai

get started

Find your pathTake Free Evaluation

community

Join the community

learn

mind: share how you thinkpathways — start freeworkshopsbootcampscoursescompare Explainxcertificationsmock testsexplainx universitycorporate traininglearn skills & mcp

discover

skillsmcp serversexplainx mcpestacktoolsmdx readeragentsllmsdesignsdictionarypeopleagi trackerfelony benchranks

company

aboutvisionmissionteaminstructorsteach on explainxpartnershipscommunityhackathonscareers

content

daily AI newsstate of AI — live resultsblogreleasespromptsgeneratorsresource libraryfor LLMsexplainx.ai kids

solutions

all solutionsdeveloper upskillingmarketing upskillingproduct manager upskillingleadership upskilling

newsletter · weekly

Get AI news, tools, and insights in your inbox.

supportcontactprivacytermsdata rightshow we create contentsubmission guidelines

© 2026 AISOLO Technologies Pvt Ltd

explainx.ai

On this page

  • Update (November 8, 2026): open RoundSquare Lean file argues mass gap and OS axioms clash on R⁴
  • Update (November 7, 2026): co-proposer Russell Impagliazzo tells the New York Times the disproof is "very shocking news"
  • Update (October 27, 2026): a second, unconfirmed claim attributes an ETH refutation to an Anthropic model
  • Update (November 11, 2026): Yang-Mills Lean code and assessment are public
  • Update (October 28, 2026): the Yang-Mills Working Group publishes its own statement
  • Update (October 27, 2026): the math advisory group asks OpenAI to release the Yang-Mills and QCD outputs now
  • Update (October 27, 2026): researchers push back on handpicked "shepherds"
  • What the Anthropic preprint changes
  • TL;DR: the three results and their status
  • What is the Exponential Time Hypothesis, and why would a refutation matter?
  • Kakeya and local smoothing: the math result released as raw output
  • The Yang-Mills and QCD claim: the one nobody can inspect yet
  • The new release process: four principles and three experiments
  • What is verified and what is not
  • What developers and readers are saying
  • What this means for what you build
  • How to check it yourself
  • Related reading
← Back to blog

explainx / blog

OpenAI Says Its Model Disproved the Exponential Time Hypothesis: What Is Claimed, What Is Verified

OpenAI, AI Research, Mathematics, Lean, AI Safety

Part of OpenAI and ChatGPT

OpenAI claims its model refuted the Exponential Time Hypothesis, proved Kakeya in all dimensions and Yang-Mills existence. What is verified so far.

Oct 26, 2026·18 min read·Yash Thakker
add explainx.ai
go deep
OpenAI Says Its Model Disproved the Exponential Time Hypothesis: What Is Claimed, What Is Verified

OpenAI said on October 26, 2026 that its internal models produced three new results: a disproof of the Exponential Time Hypothesis (ETH) in complexity theory, a proof of the local smoothing conjecture that settles the Kakeya conjecture in every dimension, and a mathematically rigorous existence proof for four-dimensional Yang-Mills theory and QCD. The ETH and harmonic analysis results come with public Lean files. The physics result does not yet, by choice of the expert group handling it.

This is a claim from the lab that built the system, not a peer-reviewed finding. The honest framing is the one OpenAI itself uses in the announcement: the manuscripts are "raw research data" that outside experts now need to turn into understanding. Below we separate what OpenAI states, what has a machine-checkable artifact, and what is still open.

Daily briefFrom Yash · 350K+ students taught

Catch up on AI

One AI story explained in plain English, plus one thing to try. Every day, free. Unsubscribe anytime.

Update (November 8, 2026): open RoundSquare Lean file argues mass gap and OS axioms clash on R⁴

Victor Logvinovich published an updated round-square repository with a long Lean 4 script and a Zenodo v5 preprint that argues the Clay conjunction on non-compact R⁴ cannot hold: strict Osterwalder–Schrader / Wightman axioms and a positive mass gap are mutually exclusive in the formal class the author defines. The README highlights an elementary Peano arithmetic contradiction on an infrared mode ansatz, and names OpenAI's October 26 physics announcement as stopping short of a public Δ greater than zero proof while artifacts remain withheld.

Independent summaries of an honesty.tools build audit (circulating in early November 2026) report that the repo compiles without extra axioms but does not yet formalize the full Clay gauge-theory statement. That is a different question from whether OpenAI's unreleased Yang-Mills materials are correct. We walk through definitions, audits, and what readers can run in a browser in RoundSquare vs OpenAI Yang-Mills.

Update (November 7, 2026): co-proposer Russell Impagliazzo tells the New York Times the disproof is "very shocking news"

Cade Metz reported for the New York Times on November 6 how complexity researchers are reacting to OpenAI's claimed ETH refutation. We read the piece through DNYUZ's republication, which carries the Times byline. The Times site blocked our fetcher with 403, so we did not open the canonical nytimes.com URL. The Hindu and Science had already quoted Impagliazzo in earlier weeks; Metz adds accessible analogies and longer quotes aimed at a general audience.

What Impagliazzo told Metz, per the republication:

  • He used Sudoku-style puzzles to explain why many computer scientists assumed solving hard problems must take far longer than checking an answer, the intuition behind both P versus NP and the Exponential Time Hypothesis he proposed in 1998 with Ramamohan Paturi and Francis Zane.
  • OpenAI's claimed result hints that machines could in theory solve very large puzzles as fast as they verify them, and could rapidly design algorithms and electrical circuits, helping computers design their own successors.
  • On circuits: "If you have a really fast algorithm for Sudoku solving, you could use it to design optimal circuits."
  • He called the disproof "very shocking news" and added, "It is also a shock that the work was not done by a human being."
  • Among theoretical scenarios the piece raises, the disproof could one day point toward faster chip design, drug discovery or breaking encryption. Metz states explicitly that these are theoretical and that immediate practical applications do not exist yet. Ryan Williams is quoted agreeing there are no immediate applications while saying the result could matter later.

The Times piece also places OpenAI's announcement inside a wave of roughly 400 claimed results from frontier labs, notes Anthropic's parallel ETH claim, and recalls that interactive proofs took decades to become practical before underpinning blockchain work. It frames recursive self-improvement as the long-run concern: Impagliazzo says the finding "could lead to better A.I., in all sorts of ways."

What this changes for verification status: nothing in the sourcing settles whether the mathematics is correct. Impagliazzo's quotes are reaction and explanation, not a peer review. The public Lean files on OpenAI's repo remain the checkable artifact; Williams, Impagliazzo and Paturi are still shepherding an open examination, as described in our table below and in Science's earlier reporting. The public openai/math repository remains the place to run the Lean toolchain against OpenAI's posted formalizations.

Update (October 27, 2026): a second, unconfirmed claim attributes an ETH refutation to an Anthropic model

On October 27, posts on X began circulating a parallel claim: that an internal Claude research model at Anthropic "independently" found a core algorithm refuting the Exponential Time Hypothesis, on October 16, using about 14 million output tokens with no human intervention, with Anthropic scientists then helping to present it in a paper. The accounts relaying it include @alpoge, @andrewcurran_ and @kimmonismus, according to an aggregator summary of the posts.

What we could and could not confirm:

  • A primary source has now appeared. A preprint titled "A 2^o(n) Time Algorithm for 3-SAT" is posted on Hexagon (version 1, October 27), listing Anthropic as contributor and submitted by Jelani Nelson on Anthropic's behalf. It says the core algorithm was found on October 16 by an internal Claude research model with no human input, that it is deterministic, that it also covers #3-SAT, and that it has been formalized in Lean 4. The provenance statement says Anthropic researchers worked with Claude to improve the presentation, and acknowledges that an internal OpenAI model claims a similar result. It carries no peer review, and we have not independently checked the proof or the Lean formalization, so the result remains a lab claim until experts weigh in.
  • The date matches OpenAI's. OpenAI's own announcement also dates its natural-language ETH disproof to October 16. Two labs reaching the same result on the same day is possible, but it is also possible the posts conflate the two stories, so wait for the paper.
  • Anthropic's documented complexity result is different. The best-supported Claude claim so far is the 3SUM and APSP algorithm, which we cover in Did Claude break the 3SUM conjecture?. That is a separate result, and it does not refute ETH.
  • What would settle it: a linked paper, a Lean file or other machine-checkable artifact, and a statement from experts such as Russell Impagliazzo, Ramamohan Paturi or Ryan Williams, who are already examining OpenAI's version.

Update (November 11, 2026): Yang-Mills Lean code and assessment are public

The Gross committee's November 9 statement promised a November 10 GitHub drop. By November 11, yang-mills-group/OpenAI-YM-20261028 is public, with an assessment PDF linked from the group page. See our standalone walkthrough: OpenAI Yang-Mills Gross group validation.

The committee still says mass gap and confinement are open. For the parallel RoundSquare thread, see our November 8 Lean post.

Update (October 28, 2026): the Yang-Mills Working Group publishes its own statement

The physics group responded in public. The Kavli Institute for Theoretical Physics homepage now carries a "Statement by the Yang-Mills Working Group", which describes David Gross, KITP permanent member and Chancellor's Chair of Theoretical Physics, as "chairing a scientific collaboration to interpret an AI result" about the existence of quantum chromodynamics and completions of the Standard Model, including a subproblem proving the existence of quantum Yang-Mills theory. What this does and does not change: the statement frames the group as an interpretation effort, not as a verdict on correctness. We could only read the homepage summary, not the full statement text, so we cannot say whether it sets a release date for the Lean files or the natural-language proof. The AI artifacts were still not public at the time of writing, and no independent check of the physics claim has been reported. The advisory group's request for immediate release, covered just below, therefore remains unanswered in the sources we could open.

Update (October 27, 2026): the math advisory group asks OpenAI to release the Yang-Mills and QCD outputs now

The Advisory Group on Mathematics and Artificial Intelligence (AGMAI), the independent panel hosted at the Institute for Advanced Study that we covered in OpenAI's math advisory group post, published a recommendation on agmai.org dated October 26 titled "Recommendation for Immediate Release of Yang-Mills AI Output." It asks OpenAI to make public, without delay, the natural-language and Lean outputs behind its claimed existence proofs for Yang-Mills theory, QCD and a completion of the Standard Model.

The group's reasoning, as stated on its page: understanding "should develop in the open, with everyone given the same access"; it expresses "great respect" for the scientists OpenAI asked to lead the physics work; and since OpenAI said it delayed release until the Lean formalization was complete, which by OpenAI's own account is now done, there is "no scientific reason for delaying the release any further."

Two points of context. First, this is the contrast already in the table below: the ETH and Kakeya materials went to GitHub, while the physics materials were held back at the request of the review group led by David Gross. Second, the AGMAI recommendation verifies no proof; it is about release process, not correctness. OpenAI had not, as far as we could find, publicly responded to the recommendation at the time of writing.

Update (October 27, 2026): researchers push back on handpicked "shepherds"

In a report for The Verge, reporter Robert Hart says researchers welcomed the smaller batch of three results, compared with the hundreds of preprints released at the start of October, but questioned the release strategy. Per The Verge, mathematics got the raw AI output, computer science gets an external group to explain the result, and physics has not been released at all.

Reactions quoted there: Constantin Kogler of the Institute for Advanced Study said he was fine with the release of the problems but that restricting first access to experts OpenAI chose "doesn't make any sense." Andreas Karch, a physics professor at the University of Texas at Austin, called the handpicked group "a disaster" and said large labs "should not be picking winners and losers." Brown mathematician Brendan Hassett said the Kakeya claim "looks well supported" but that he had not checked the Lean formalization. Stanford's Jared Duker Lichtman said that with any approach "there is going to be someone who is upset."

The Verge also notes that AGMAI had recommended a day earlier that AI companies not adopt the shepherd practice, favoring scholarly repositories, formalized proofs and funded support. As above, none of this verifies a proof; these are expert opinions on process, reported by one outlet.

What the Anthropic preprint changes

The Hexagon preprint moves the Anthropic claim from social-media rumor to a lab claim with a primary document, the same status as OpenAI's. Both labs now say an internal model reached a sub-exponential 3-SAT result around October 16, and the preprint itself acknowledges the OpenAI claim. Neither has been peer reviewed, and we have not verified either proof or Lean formalization.

We will update this section as experts respond.

TL;DR: the three results and their status

table · 5 cols
ResultFieldOpenAI's claimPublic artifact todayWho is shepherding
Local smoothing and KakeyaHarmonic analysisLocal smoothing proved in every dimension, implying Kakeya and Stein's restriction conjecture in all dimensionsAI write-up plus Lean proof on OpenAI's GitHubReleased as raw output; researchers organize their own efforts
Exponential Time Hypothesis is falseComplexity theorySystem refuted ETH, verified in LeanAI write-up plus Lean proof, all raw data to be publicRussell Impagliazzo, Ramamohan Paturi, Ryan Williams
Yang-Mills and QCD existenceQuantum field theoryRigorous existence proof of certain 4D gauge-plus-fermion theoriesNot released; group holds AI artifactsDavid Gross, Michael Douglas, Arthur Jaffe, Igor Klebanov, Nikita Nekrasov

The three different release strategies are themselves the news. OpenAI says outside experts advised each one, and that it does not yet know which will work best.

What is the Exponential Time Hypothesis, and why would a refutation matter?

The ETH is a conjecture about how hard Boolean formula satisfiability is, a problem closely tied to P versus NP. In OpenAI's words it "has been widely believed to be true, and has been used in hundreds of papers to derive conditional computational hardness results" across graph theory, computational biology, databases and computational geometry.

That last part is why this would matter if it holds. Many papers say "this problem cannot be solved much faster, unless ETH is false." If ETH is false, those conditional lower bounds lose their foundation, and researchers get a reason to hunt for faster algorithms on problems they had written off.

It is not a claim that P equals NP, and OpenAI does not make one. A Hacker News commenter asked exactly that, wondering whether the news should shift priors toward P=NP (HN comment); the announcement itself only says the refutation "reopens questions about how much faster these tasks could become" and that practical gains "will require significant further research."

The stated timeline: the natural-language disproof was found on October 16, 2026 and formalized in Lean on October 17. The collaboration group is led by Impagliazzo and Paturi, who with Francis Zane originally proposed ETH, and Williams, who has previously refuted stronger versions of it. They asked that all raw data be made public for independent verification, and OpenAI agreed, plus a commitment to fund a conference on the implications. The group has decided to publish the natural-language proof and Lean formalization immediately.

For background on how we have treated similar complexity claims, see our fact-check of the Claude 3SUM and APSP claim and the integer multiplication Lean proof post.

Kakeya and local smoothing: the math result released as raw output

The mathematics result concerns waves. Sogge's 1991 local smoothing conjecture says the most extreme focusing of waves must be short-lived: the amount of focusing at a typical time is much smaller than the worst case. It is a foundational statement, and it implies the Kakeya conjecture and Stein's restriction conjecture.

OpenAI lays out the prior record. Guth, Wang and Zhang proved local smoothing in two spatial dimensions in 2019. Wang and Zahl proved the Kakeya conjecture in three dimensions in 2025. OpenAI says its models proved local smoothing in every dimension, and therefore Kakeya and Stein's restriction conjecture in all dimensions, via "a novel auxiliary partial differential equation."

Timeline per OpenAI: AI output on October 18, Lean formalization on October 19, then an outreach to an expert, who recommended publishing the AI output and letting researchers organize themselves, the same strategy as the October 6 release. Both the natural-language write-up and Lean proof are in the public GitHub repository.

Readers should keep the earlier release in mind. The October 6 release of 722 manuscripts came with Lean formalizations for many, but not all, proofs, and OpenAI said the average result used roughly three hours of ChatGPT Pro thinking. Our breakdown of eight headline results tracks which of them have Lean status. A Kakeya-in-all-dimensions result, if it stands, would be of a different order than most of those.

The Yang-Mills and QCD claim: the one nobody can inspect yet

OpenAI describes "a mathematically rigorous existence proof of certain interacting quantum field theories containing gauge fields and fermions in four dimensions," including QCD and a unified completion of the Standard Model, building on Balaban's constructive approach. As a subproblem, it includes existence of pure Yang-Mills theory, which is part of the Clay Mathematics Institute Millennium Prize problem on Quantum Yang-Mills Theory.

Timeline per OpenAI: natural-language Yang-Mills proof on September 9, 2026; QCD proof on September 18; Lean formalization of Yang-Mills existence finished October 16. The company says it delayed release because of "the length and complexity of the work" until the Lean formalization completed.

The shepherding group is led by David Gross, with Michael Douglas, Arthur Jaffe, Igor Klebanov and Nikita Nekrasov. Gross shared the 2004 Nobel Prize for asymptotic freedom, and Jaffe co-formulated the Millennium problem with Witten. The group "decided to hold off on releasing the AI-produced artifacts and instead work on writing it up themselves," and asked that communications go to a dedicated email address.

Our standing guidance on Millennium claims applies. Before this announcement, our fact-check of the seven Millennium Prize Problems concluded nothing on the list had been confirmed as solved by AI. Until the group publishes its own write-up, this result is an unverifiable-by-the-public claim from OpenAI, backed by named experts who have agreed to examine it, not by a statement that they have confirmed it. Note that the announcement says the group will "understand and explain" the result; it does not say they have validated it.

The new release process: four principles and three experiments

The least flashy part of the announcement may age best. OpenAI says its "release process continues to evolve in response to feedback from the associated academic communities." The four principles for the external collaborations:

  • Transparency: collaborations are disclosed, no nondisclosure agreements, participants can discuss and publish freely.
  • No academic credit: OpenAI says it does not seek credit for discoveries produced by its internal models, and inviting a researcher to steward a group does not imply they made the discovery.
  • Recognition: the group gets recognition for the understanding and write-ups it produces, and helps ensure prior work is credited properly.
  • Independence: participation is voluntary and unpaid, and members do not speak for OpenAI.

OpenAI also names the weakness itself: "choosing initial shepherds necessarily shapes who gets involved first and creates asymmetry of information." That echoes criticism we covered, including the Association for Human Mathematics statement and the AGMAI checklist for responsible release of AI mathematics. The company also points to its $5 million contribution to a Simons Foundation initiative, covered in our Simons Foundation $15M post.

What is verified and what is not

table · 2 cols
QuestionStatus as of October 26, 2026
Did OpenAI publish the ETH write-up and Lean proof?Yes, per OpenAI, in its public GitHub repository
Has an independent expert confirmed the ETH disproof?Not yet; the named group has just been convened to examine it
Does the Lean proof prove the statement people mean by ETH?Needs a human check that the formal statement matches the conjecture
Is the Kakeya-in-all-dimensions proof public?Yes, write-up and Lean file; no independent report seen yet
Is the Yang-Mills proof public?No; the expert group is holding the AI artifacts
Has anyone outside OpenAI confirmed Yang-Mills existence?We found no such statement

A Lean check is strong evidence that a formal statement follows from its axioms, but not that the formal statement captures the informal claim. Our Lean 4 proof cost collapse explainer covers why that matters, and our earlier look at OpenAI's Navier-Stokes agent swarm claim shows how a claim can outrun its verification. "Checked in Lean" is a pointer to where scrutiny should go, not the end of it.

What developers and readers are saying

The Hacker News thread was small when we read it, with eight comments. One commenter, rbranson, noted that the post title buried the most striking claim, that the ETH is false (comment). Another, mlinsey, asked whether complexity theorists now consider P=NP more likely (comment). Treat these as early reactions from commenters, not expert verdicts. The expert groups named by OpenAI have not yet published analysis.

What this means for what you build

Most readers will not touch Kakeya or Yang-Mills directly. The practical stakes are elsewhere:

  1. Verification is becoming the bottleneck. If a frontier model can emit a Lean-checked proof in a day, the scarce resource is expert time to read it. Teams using AI for proofs, protocol verification or smart-contract analysis should budget for review of the formal statement, not just a green check.
  2. Conditional hardness results may need re-reading. If your work cites ETH to justify why something cannot be sped up, track the expert group's write-up before changing anything.
  3. Release practice is being invented live. Three strategies in one announcement is a useful case study if you publish AI-generated research or code and need to decide who gets access first and who gets credit.
  4. Expect more. OpenAI says it is "working to responsibly release the model" behind the October 6 results, and our advisory group coverage shows how quickly this has escalated since September.

How to check it yourself

  1. Read the OpenAI announcement and the October 6 release post, which explains the repository layout and protocols for paper revisions and citations.
  2. Open the openai/math repository, read the natural-language write-ups, and note which results have Lean files.
  3. Install Lean 4 and build the proofs. Then read the final theorem statement and ask whether it says what the headline says.
  4. Wait for the named groups. For ETH, that means Impagliazzo, Paturi and Williams; the announcement says they invite an open examination and plan a funded conference.
  5. Watch for dissent. Our expert reactions post shows that earlier releases drew mixed responses from mathematicians.

Related reading

  • Follow-up: the AI factoring claim, what Lean checks

  • OpenAI's 722 math manuscripts: what the repo contains

  • Eight headline results and their Lean status

  • The seven Millennium Prize Problems: what AI has actually solved

  • Did Claude break the 3SUM conjecture? Claimed versus verified

  • AGMAI's rules for releasing AI math

  • Simons Foundation, OpenAI and Anthropic launch a $15M effort

Details are accurate as of October 26, 2026, the day of OpenAI's announcement. Expert analysis of these results is just beginning, so check the linked sources for updates.

Related — October 28, 2026: a viral thread says games and simulators explain these results; we checked each claim.

Spotted something out of date? Let us know.
Daily briefFrom Yash · 350K+ students taught

Catch up on AI

One AI story explained in plain English, plus one thing to try. Every day, free. Unsubscribe anytime.

  • ✓ Free
  • ✓ One email a day, no spam
  • ✓ Unsubscribe anytime
Yash Thakker

Written by

Yash Thakker

Yash is an AI expert with over 300K learners. Join his workshops →

View Yash Thakker in People in AI →

Related posts

Oct 6, 2026

OpenAI Releases 722 Math Manuscripts: What the openai/math Repo Contains

On October 6, 2026, OpenAI put 722 mathematical manuscripts from an unreleased internal model into a public GitHub repository, openai/math. This post explains what is in it, how the roughly three-hours-of-ChatGPT-Pro-per-result figure should be read, what Lean does and does not prove, and how to triage a release this large.

Nov 11, 2026

OpenAI Yang-Mills Lean Proof: Gross Group Validation and Public Release

The Yang-Mills Working Group chaired by David Gross published OpenAI's October 28 Lean submission on GitHub and an expert assessment. The committee says the formalization proves the existence of pure Yang-Mills theory under standard statements. Confinement and the mass gap remain unresolved.

Oct 17, 2026

What OpenAI's 700+ Math Results Leave Out: Optimization, ML Theory and Cryptography

After OpenAI released more than 700 AI-proved math results, researcher Daniel Paleka compared the fields covered against arXiv and found statistics, ML theory, optimization and numerical analysis nearly absent. This post explains the finding, the cryptography speculation around it, and what the data cannot prove.