Will any AI lab solve a Millennium Prize Problem before 2027?

- [2026-09-08] OpenAI announced an unreleased internal model (a swarm of ~10,000 AI agents) took 88 hours to produce the solution, deploying as many as 10,000 AI agents working in parallel , claiming a solution to the Navier-Stokes existence/smoothness Millennium Prize Problem (https://tribune.com.pk/story/2628177/openai-says-ai-model-solved-one-of-mathematics-millennium-prize-problems). - [2026-09-08] The claimed result addresses only the "forced" alternatives (Fefferman's statements C and D), not the original unforced/zero-force case; the Navier-Stokes existence and smoothness problem with a real-analytic external force or no external force remains unsolved (https://en.wikipedia.org/wiki/Navier%E2%80%93Stokes_existence_and_smoothness). - [2026-09-08] In the Tuesday announcement, OpenAI said the company does not intend to claim the Millennium Prize for the result. (https://www.cnn.com/2026/09/09/business/openai-millennium-problems-navier-stokes-hnk). - [2026-09-08] A priority/credit dispute erupted: the night before, NYU's Tristan Buckmaster and Anthropic researcher Levent Alpöge had been looking into the Navier-Stokes problem –– a culmination of work that began last year – and had used a mix of large language models, including OpenAI's Codex (https://www.cnn.com/2026/09/09/business/openai-millennium-problems-navier-stokes-hnk). - [2026-09-10] OpenAI updated its account, with Kingy reporting OpenAI says verification preceded the September 6 discussions... After an internal investigation, OpenAI now says Buckmaster's Codex prompts from the preceding two months could not have influenced the system, including through training (https://kingy.ai/blog/navier-stokes-ai-proof-claims-dispute/). - [2026-09-11] The Clay Mathematics Institute issued a statement that "the Navier–Stokes problem has apparently been settled" but maintained the process for verification and credit assignment "is deliberately unhurried"; the status of the Navier–Stokes problem is now "active" instead of unsolved or solved on the Institute's website (https://en.wikipedia.org/wiki/Navier%E2%80%93Stokes_priority_controversy). - [2026-09-11] CMI president Martin Bridson called the series of announcements "exciting", he stressed that the Institute requires a solution to a Millennium problem to be in the form of peer-reviewed publications, which would then be further scrutinised before the prize is awarded (https://en.wikipedia.org/wiki/Navier%E2%80%93Stokes_priority_controversy). - [2026-09-19] (per 10-day-old report) CMI's response described as triggering the Navier-Stokes existence and smoothness problem, one of the Millennium Prize Problems, could have been settled. This now triggers a formal, multi-year verification process (https://www.neowin.net/news/openai-agent-swarm-triggers-verification-for-navier-stokes-math-problem/). - [2026-09-29] A later report reaffirms that The Clay Mathematics Institute has not accepted the result and still lists the problem as unsolved, and the NYU mathematician whose related work came first says OpenAI took that route only after hearing about it (https://www.implicator.ai/clay-institute-navier-stokes-openai-proof-claim/). - [2026-09-10] A human-audit arXiv paper of OpenAI's broader set of ten August 2026 math results found No surviving confirmed substantive mathematical error in a principal result appears in the assessments examined , suggesting no retraction so far, though this predates/parallels the Navier-Stokes-specific claim (https://arxiv.org/pdf/2608.14673). - [2026-08-10] Separately, an unreleased Anthropic Claude model improved a bound related to the Riemann Hypothesis (zeros on the critical line from 41.6% to 67.2%) but explicitly did not solve that Millennium Problem; Claude didn't prove the hypothesis. But somewhere in the attempt, working on a side question that falls out of the same territory, Claude landed a genuine improvement on a long-standing lower bound (https://www.datacamp.com/tutorial/claude-and-the-riemann-hypothesis). - [2026-09-30] As of the latest Wikipedia update (cited as "0 days ago"), the matter remains an active, unresolved dispute with no CMI certification, peer-reviewed publication, or prize award yet, and the problem is listed as "active" rather than solved (https://en.wikipedia.org/wiki/Navier%E2%80%93Stokes_priority_controversy). BOTTOM LINE: A major, credible, and widely reported AI claim (OpenAI, Sept 8, 2026) toward solving the Navier-Stokes Millennium Prize Problem exists and has not been retracted, but as of Oct 1, 2026 it remains unverified by CMI, unpublished in a peer-reviewed journal, and formally still "active"/unsolved per CMI's own site, with a credit dispute ongoing — meaning the strict resolution criteria (CMI/journal/two-agency confirmation) have not yet been met, though the process is actively underway with 3 months left before close. NOT FOUND: No evidence of CMI or peer-reviewed-journal certification of the OpenAI Navier-Stokes result as of today; no evidence of a retraction/error finding that would void the claim; no evidence any other Millennium Problem (P vs NP, Hodge, BSD, Yang-Mills) has been claimed solved by an AI lab in this window.

- [2026-09-08] Before this year, in a quarter-century since 2000, only one has been solved — the Poincaré Conjecture — solved by Russian mathematician Grigori Perelman in 2006 , giving a historical base rate of roughly 1 solution per 25 years (0.04/yr) prior to the AI era. https://www.quantamagazine.org/ai-has-solved-one-of-maths-1-million-millennium-prize-problems-20260908/ - [2026-09-08] That base rate was broken this year: OpenAI shared an AI-generated solution to the Navier–Stokes Millennium Prize Problem, including a writeup and a formal proof in Lean, produced by an internal OpenAI system, showing that the dynamics of the equations can develop a singularity in finite time , per OpenAI's own announcement. https://openai.com/index/navier-stokes-solution/ - [2026-09-08] Independent reporting confirms the scale of the effort: OpenAI deployed around 10,000 agents of an as-yet unreleased powerful model on Aug. 28, and solved it within 88 hours, proving that the equations sometimes break down under extreme conditions . https://www.semafor.com/article/09/08/2026/openai-agents-find-proof-to-1-million-millennium-prize-problem - [2026-09-08] This is the key precedent for "another" solution: the current question requires a *second, distinct* problem solved by Jan 1 2027, since Navier-Stokes (this question's predecessor event) was already claimed before this market's Sept 12 issuance. Wikipedia already reflects the update: in September 2026, the American artificial intelligence company OpenAI presented a proposed counterexample to the Navier–Stokes existence and smoothness problem, saying that they do not intend to claim the Millennium Prize for the result . https://en.wikipedia.org/wiki/Millennium_Prize_Problems - [2026-04-xx] Base-rate expectations shifted dramatically in months: as recently as April this year, prediction markets gave AI less than a 40% chance of solving any Millennium Prize Problem before 2030 — yet one fell within the same year. https://thezvi.wordpress.com/2026/09/13/brand-new-ai-solves-a-millennium-prize/ - [2026-09-11] Signals of an imminent second result exist but are unconfirmed: OpenAI told the New York Times it has made "substantial progress on another Millennium Prize problem" since completing Navier-Stokes, confirmed as progress on a second, unnamed problem, but not confirmed which problem, with rumors about Hodge or Birch–Swinnerton-Dyer unverified . https://explainx.ai/blog/millennium-prize-problems-ai-solved-fact-check-2026 - [2026-08-xx] A separate, partially-related claim exists on the Riemann Hypothesis: Anthropic reported in August 2026 that an unreleased research version of Claude improved a long-standing bound tied to the Riemann Hypothesis, raising the proven lower bound for the fraction of Riemann zeta zeros on the critical line — this is a partial result, not a full solution, and explicitly would not satisfy this contract's "complete proof" requirement. - [2026-09-xx, sibling market] A directly analogous Polymarket contract on "how many more" Millennium Prize Problems (excluding Navier-Stokes/Poincaré) AI will solve in 2026 prices **0 additional solutions at 61%**, 1 at 28%, 2+ at ~11% combined, reflecting trader skepticism that a second full solution lands within the year. https://polymarket.com/event/how-many-more-millennium-prize-problems-will-ai-solve-in-2026 - [2026-09-09, sibling market] A related Polymarket contract "OpenAI announces another Millennium Prize solution by...?" prices the Sept 30 2026 date at only ~2%, the Dec 31 2026 date at ~36%, and the Dec 31 2027 date at ~70% — implying most trader mass for a second OpenAI solution sits beyond this year, not before it. https://polymarket.com/event/openai-announces-another-millennium-prize-solution-by - [2026-xx-xx] Historically, verification of Millennium-level proofs has taken years, not weeks: verifying Perelman's proof took teams of mathematicians several years and hundreds of additional pages of exposition , suggesting that even if an AI-authored proof is announced before Jan 1 2027, the "unretracted and officially recognized" bar in this contract's rules could be hard to clear that fast — though the Navier-Stokes case shows Lean-formalization can compress this timeline considerably. - [2026-09-08] There is open controversy/priority dispute attached to the one confirmed case, which is relevant to how fast future claims might be accepted or retracted: the massive result is not without controversy; the proof of a "blowup" has blown up the mathematics community , and a competing human/AI-assisted team (Buckmaster/Alpöge) disputed OpenAI's independence and priority. https://www.quantamagazine.org/ai-has-solved-one-of-maths-1-million-millennium-prize-problems-20260908/ BOTTOM LINE: The historical base rate for any Millennium Prize solution was extremely low (≈1 per 25 years) until 2026, when OpenAI's Navier-Stokes result broke that pattern entirely — but that prior case does not count for this "another/second" question, and both sibling prediction markets (Polymarket's "how many more in 2026" and "OpenAI announces another solution by…") currently assign a minority probability (roughly 30-40%) to a second full solution materializing before year-end 2026, with more mass on 2027+, despite credible reports of "substantial progress" on an unnamed second problem. NOT FOUND: No confirmed identification of which specific second Millennium Prize Problem OpenAI (or any other lab) is reportedly close to solving; no evidence of any second problem having reached the "unretracted, officially recognized" bar as of early October 2026; no official CMI statement recognizing either the Navier-Stokes claim or any other claim as a verified solution.

YES means: a second (or first fully-qualifying) Millennium Prize Problem is "solved" under the contract's strict criteria — CMI/journal certification or two Source Agencies confirming a specific unretracted written work — by an AI lab, between issuance and Jan 1, 2027 (the Navier–Stokes claim alone does not count toward "another" problem unless it is itself the qualifying solved problem and the rules treat it as counting). - [2026-09-17] OpenAI itself confirmed to the New York Times — its most authoritative public statement to date — that since announcing on September 8 that it had solved the Navier-Stokes equations, it has made "substantial progress" on a second Millennium Prize Problem—but did not disclose which problem or provide a timeline for disclosure (https://www.kucoin.com/news/flash/openai-reports-major-progress-on-second-millennium-prize-problem). - [2026-09-17] The Information, citing a single internal source, reported that employees at the AI firm expect to soon crack the Hodge Conjecture , making Hodge the leading (though unconfirmed) candidate for the "second problem" (https://gizmodo.com/openai-reportedly-trying-to-solve-hodge-conjecture-amid-feud-with-math-community-2000813658). - [2026-09-17] The same reporting indicates OpenAI is deliberately slowing its rollout this time: OpenAI has not announced a solution to the Hodge Conjecture, but employees reportedly expect one soon and the company may consult mathematicians before publishing it , explicitly to avoid repeating the Navier–Stokes PR/credit fiasco (https://superpowerdaily.com/posts/openai-may-delay-a-reported-hodge-conjecture-result-after-math-backlash). - [2026-09-17] A detailed sourced account notes OpenAI's Navier-Stokes submission already demonstrates the mechanism a second solution would likely follow: OpenAI has not announced a second Millennium Prize solution... Employees expect a solution soon, according to a person familiar with the matter , with the same 10,000-agent swarm/Lean-formalization pipeline potentially reusable for Hodge (https://kingy.ai/news/openai-next-millennium-prize-problem-hodge/; https://the-decoder.com/openai-reportedly-closes-in-on-solving-the-hodge-conjecture-its-second-millennium-prize-problem/). - [2026-09-10ish] Rumor-stage reporting (unverified) also names Birch–Swinnerton-Dyer as a parallel target, with claims that OpenAI and its rival Anthropic appear to be in a wildly intense secret race, both launching their final push toward the third great challenge, the "BSD Conjecture," with some work already entering the verification stage — giving the YES path two live candidate problems, not just one, raising surface area for a win before January (https://www.kucoin.com/news/flash/openai-claims-breakthroughs-on-multiple-millennium-math-problems). - [2026-09-17] Crucially, the capability bar for a second claim is now demonstrated, not hypothetical: OpenAI's Navier-Stokes submission already used OpenAI’s own statement that “Substantial progress” on a second, unnamed Millennium Prize problem as a template, and per Kingy AI's sourced breakdown the original effort involved roughly 10,000 concurrent agents, 2.7 million messages, and around 130 billion output tokens, completed in about 88 hours plus 17 hours of Lean formalization — a rapid, repeatable production process a lab could re-run on Hodge or BSD within weeks (https://kingy.ai/news/openai-next-millennium-prize-problem-hodge/). - [2026-09-11] The CMI's own posture toward Navier-Stokes — the Clay Mathematics Institute said that the problem had "apparently been settled," while stressing that its process for evaluating what had been achieved and assigning credit would be "deliberately unhurried" — shows that even an imperfect or disputed AI proof is being treated seriously by the gatekeeper body, lowering the bar psychologically for the next claim to receive similar high-level attention quickly (https://vplevris.medium.com/is-hodge-next-openai-claims-progress-on-a-second-millennium-problem-b4270bdc2dac). - [2026-09-22] Nonetheless, outside analysis stresses the CMI's formal process is slow and journal-publication-gated: The Clay Mathematics Institute's published rules require a proposed solution to appear in a refereed mathematics journal of worldwide repute. Then two full years of general acceptance in the mathematics community must follow that publication — a timeline structurally incompatible with a Jan 1, 2027 deadline unless the contract's "two Source Agencies report" clause (an alternative to CMI/journal certification) is triggered instead (https://fourweekmba.com/ai-openai-hodge-conjecture-clay-scoreboard/). - [2026-09-29] As of the most recent dedicated tracking, no named conjecture, exact theorem, public proof, or formal artifact for a second problem has emerged: as of writing, there is no paper, no code, and no statement from the Clay Mathematics Institute publicly available on the Hodge Conjecture (https://fourweekmba.com/ai-openai-hodge-conjecture-clay-scoreboard/). - [2026-09-xx] A sibling Manifold market tracking this exact dynamic ("Did OpenAI just solve Hodge??") frames the resolution bar precisely as this contract does, noting Resolves yes if OpenAI claims a proof of Hodge's conjecture which is seen by consensus as correct by market deadline and that as of its Oct 1 deadline commentary, OpenAI has not even named the problem yet — reinforcing that the "two Source Agencies confirm a specific written work" bar for YES remains unmet to date (https://manifold.markets/Incompleteusern/did-openai-just-solve-hodge). BOTTOM LINE: The strongest YES case rests on OpenAI's own September 17 admission of "substantial progress" on an unnamed second Millennium Prize Problem (widely but not officially attributed to Hodge, with Birch–Swinnerton-Dyer also rumored), combined with a demonstrated, fast, repeatable agent-swarm-plus-Lean-formalization pipeline that produced the first (Navier-Stokes) claim in under four days — meaning a second full announcement is technically plausible well before Jan 1, 2027, even though no named problem, proof, or artifact has surfaced yet as of late September 2026. NOT FOUND: No named second problem, no public proof/preprint, no Lean formalization, and no CMI or Source Agency confirmation of any second (or BSD) solution as of the latest fetched sources (late September 2026); no evidence of an announcement timeline from OpenAI or Anthropic.

YES means: Between 2026-09-12 (issuance) and 2027-01-01 (close), a Millennium Prize Problem is "solved" under the contract's strict criteria — i.e., unretracted by Expiration and either (a) published in a peer-reviewed journal, (b) officially recognized by CMI or another academic governing body by reference to a specific written work, or (c) reported as resolved by at least two Source Agencies citing a specific written work — and that solution is credited to an AI lab. Given CMI's own rules require journal publication plus roughly two years of community vetting before even considering recognition, and the Navier-Stokes claim is explicitly excluded as a "before time period" predecessor, YES realistically requires a second, distinct Millennium Problem to clear this very high bar in under four months. - [2026-09-10] Clay's own published rules for recognizing a solution require a **published** result in a qualifying outlet, **at least two years** to pass after publication, and general acceptance by the global math community before CMI will even consider it — a rule explicitly confirmed by a source quoting Clay's rules page directly, making any under-four-month "official recognition" of a brand-new claim procedurally almost impossible under CMI's own stated process (https://manlius.substack.com/p/no-ai-hasnt-officially-solved-navierstokes). - [2026-09-12] A dedicated review of the Navier-Stokes claim states the claim status is "Not yet verified," the review process is "Expected to take multiple years," and CMI's evaluation is explicitly described as "intentionally conservative" (https://www.digitalcitizen.life/openai-navier-stokes-claim-enters-formal-verification-as-millennium-prize-review-begins/). - [2026-09-19] A specific independent arXiv technical paper examining the OpenAI construction confirms it only resolves Fefferman's forced alternatives (C/D), the unforced case is "untouched," and that "as of 12 September 2026 there was no refereed independent verification" — underscoring that even the predecessor claim remains scientifically incomplete relative to the original problem statement (https://arxiv.org/pdf/2609.17642). - [2026-09-22] Reporting via The Information indicates OpenAI's "primary concern is not solving the Hodge Conjecture but announcing they did so without further infuriating mathematicians," implying a deliberate internal slowdown in disclosure timing precisely because of the backlash from the Navier-Stokes episode (https://gizmodo.com/a-mathematician-discovered-a-new-eight-faced-shape-it-looks-like-a-glitch-in-reality-2000818219). - [2026-09-22] A dedicated report states that critics "reportedly signed open letters, and OpenAI withdrew sponsorship from an upcoming Caltech Mathathon," evidence of concrete organizational and reputational friction that plausibly delays any rushed second announcement (https://superpowerdaily.com/posts/openai-may-delay-a-reported-hodge-conjecture-result-after-math-backlash). - [2026-09-17] Even OpenAI's own internal framing, per The Information, is that it is merely "close to solving the Hodge Conjecture" — not that it has solved it — and this remains sourced to a single unnamed person, not an official OpenAI statement, with no paper, code, or Lean formalization released (https://kingy.ai/news/openai-next-millennium-prize-problem-hodge/). - [2026-09-22] A separate aggregator piece similarly concludes "the Hodge conjecture has only internet rumors... no corresponding paper, no code, and no confirmation from any authoritative institution" as of its date (https://vocal.media/fyi/the-hodge-conjecture-rumor-openai-a-viral-tweet-and-the-math-worlds-quiet-fury). - [2026-09-29] CMI's institutional stance has not moved in three weeks: it "still lists Navier-Stokes among its unsolved Millennium Prize problems," and the NYU priority dispute remains unresolved, suggesting CMI is unlikely to fast-track a second, more abstract (and less publicly scrutinized) claim like Hodge before year-end (https://www.implicator.ai/clay-institute-navier-stokes-openai-proof-claim/). - [2026-09-27] A survey piece of "Four forecasts" notes that even optimistic observers peg Polymarket's implied probability of an AI lab publicly claiming a second Millennium solution by year-end at only 36%, versus 70% by the following year — i.e., even a bare *claim* (let alone CMI/peer-review-level resolution) is seen as more likely to land in 2027 than in the remaining three months of 2026 (https://danjwu.substack.com/p/what-math-survives). - [2026-09-14] Social-media rumors of an imminent Hodge announcement ("this week or next") from mid-September had already proven premature by month's end, with no confirming paper, code, or institutional statement materializing three weeks later — a pattern of hype outrunning verified fact that has repeated at least twice in this saga (Navier-Stokes then Hodge) (https://x.com/imjustnewatai/status/2099632027184017633). - [2026-09-10] Historical precedent for the *only* prior solved Millennium Problem shows verification alone (not even discovery) took years: Perelman posted his decisive preprints in 2002-2003, and "the mathematical community then spent years checking and reconstructing the argument before Clay awarded him the Millennium Prize in 2010," a seven-to-eight year gap between claim and official recognition (https://manlius.substack.com/p/no-ai-hasnt-officially-solved-navierstokes). BOTTOM LINE: Under the contract's literal, high-bar resolution criteria (peer-reviewed publication, explicit CMI/academic-body recognition, or two-Source-Agency confirmation of a *specific written work*), no second Millennium Problem has even been publicly claimed to be solved as of October 1, 2026 — only unconfirmed rumor/sourcing around Hodge — and CMI's own stated process plus the Navier-Stokes precedent both point to multi-year, not multi-month, timelines for any claim to clear the bar, making a clean YES before January 1, 2027 structurally unlikely absent an extraordinary acceleration. NOT FOUND: No official OpenAI, Anthropic, or other AI lab paper, Lean formalization, or press release naming or substantiating a solved second Millennium Problem (Hodge, BSD, P vs NP, Yang-Mills, or Riemann) as of the dates searched; no CMI statement on any problem beyond its "active" Navier-Stokes status; no evidence of peer-reviewed journal publication for either Navier-Stokes or any other problem.