Will OpenAI announce another Millennium Prize solution by December 31, 2027?

- [2026-09-08] OpenAI publicly announced that an unreleased internal AI model produced a proof for the Navier–Stokes existence and smoothness problem, releasing a paper and Lean formalization; OpenAI said it does not intend to claim the Clay Institute's $1 million prize. This problem is explicitly excluded from qualifying under the market's rules. https://openai.com/index/navier-stokes-solution/ - [2026-09-09] Mathematician Terence Tao and others raised concerns about priority and the "frenetic competition" dynamic, and CNN confirmed only one Millennium Prize problem has been officially solved so far: the Poincaré Conjecture . https://www.cnn.com/2026/09/09/business/openai-millennium-problems-navier-stokes-hnk - [2026-09-10/11] Reports (Andrew Curran, MTS) said OpenAI is using its post-Navier-Stokes internal model to attempt the Riemann Hypothesis and P versus NP, among unnamed "high impact problems," but no results were disclosed. https://huggingnews.com/ai/openai-targets-riemann-hypothesis-and-p-vs-np-with-navier-stokes-model-ff5bf602 - [2026-09-11] OpenAI told the New York Times it had made "substantial progress" on a second Millennium Prize Problem since solving Navier-Stokes on September 8, without naming the problem or a disclosure timeline . https://decrypt.co/378551/openai-progress-second-millennium-prize-math-problem - [2026-09-17] The Information reported, citing a single anonymous source, that OpenAI employees expect to solve the Hodge Conjecture "relatively soon," but this is reported speculation, not an OpenAI confirmation. OpenAI is reportedly close to solving another of math's seven famous Millennium Prize Problems, working on the Hodge conjecture, with employees expecting a solution soon, according to a person familiar with the matter . https://the-decoder.com/openai-reportedly-closes-in-on-solving-the-hodge-conjecture-its-second-millennium-prize-problem/ - [2026-09-19] The Information reiterated on X: "OpenAI is close to solving the Hodge Conjecture, another of the seven Millennium Prize Problems, according to a person with knowledge of the work." Still not an OpenAI statement. https://x.com/theinformation/status/2101400971599282524 - [2026-09-21] OpenAI announced a new independent "Advisory Group on Mathematics and Artificial Intelligence" hosted at the Institute for Advanced Study, Princeton, stating the group is meant to give mathematicians more input into the company's math-oriented research, serving "as a bridge to the mathematical community and broader public." https://techcrunch.com/2026/09/21/openai-forms-math-advisory-group-as-its-ai-resolves-more-than-100-open-problems/ - [2026-09-21] In the same announcement, OpenAI said an internal model had solved more than 100 long-standing problems across most areas of mathematics on top of Navier-Stokes, after training that began August 28 — about a month — with its own mathematicians "surprised" by the pace; among the solved problems is reportedly a second Millennium Prize Problem, the Hodge conjecture. This aggregate claim has not been independently verified, with no public list, proofs, or resolution criteria released. https://tech-insider.org/openai-100-math-problems-solved-24-days-2026/ - [2026-09-22] Gizmodo reported, via The Information, that OpenAI is allegedly homing in on another Millennium Prize Problem, with employees at the AI firm expecting to soon crack the Hodge Conjecture , but also that OpenAI's "primary concern" is managing PR fallout, not the math itself. https://gizmodo.com/openai-reportedly-trying-to-solve-hodge-conjecture-amid-feud-with-math-community-2000813658 - [2026-10-04] A fact-check roundup confirmed that as of this date the public record supports only that OpenAI says it made substantial progress on an unnamed Millennium problem; claims that Hodge is already solved, that Birch and Swinnerton-Dyer is also near completion, or that an internal model called "Aeon" is responsible remain unconfirmed . https://kingy.ai/news/openai-next-millennium-prize-problem-hodge/ - [2026-10-04] Wikipedia's Millennium Prize Problems page, reflecting the latest consensus, states as of 2026 the only Millennium Prize problem officially declared solved is the Poincaré conjecture, and OpenAI's Navier–Stokes claim has not been independently verified as of 2026 ; no mention of any other qualifying problem being announced as solved. https://en.wikipedia.org/wiki/Millennium_Prize_Problems - [2026-10-04] IEEE Spectrum noted the ongoing "reckoning" among mathematicians, confirming since the Navier-Stokes announcement, rumors have swirled about which of the remaining Millennium Problems will be next, and OpenAI is reportedly focusing on solving the Hodge conjecture — still framed as a rumor/report, not a formal OpenAI claim of solution. https://spectrum.ieee.org/millennium-prize-ai BOTTOM LINE: The only concrete OpenAI Millennium Prize announcement to date (Navier-Stokes, Sept 8 2026) does not qualify under this market's rules, and as of October 6, 2026 OpenAI has only acknowledged unnamed "substantial progress" on a second problem plus an unverified "100+ problems solved" claim that reportedly includes Hodge — with no formal, named announcement from OpenAI itself declaring any of the five qualifying problems (Riemann, P vs NP, Yang-Mills, Hodge, BSD) solved. NOT FOUND: No official OpenAI blog post, paper, or statement naming the Hodge Conjecture (or any other qualifying problem) as solved; no Clay Mathematics Institute confirmation or independent mathematician verification of any qualifying-problem claim; no evidence of work on Yang-Mills; no post-October 4, 2026 updates found.

- [2026-09-09] Reference class/history: in the ~26 years since the Clay Institute established the seven Millennium Prize Problems in 2000, only one has ever been officially confirmed solved — the only Millennium Prize problem to have been officially declared solved is the Poincaré conjecture, with the Clay Institute awarding the monetary prize to Grigori Perelman in 2010, though he declined the award . That is a base rate of roughly one confirmed solution per 26 years across all solvers (human or AI), illustrating how rare official "solved" declarations are. https://en.wikipedia.org/wiki/Millennium_Prize_Problems - [2026-09-08/09] OpenAI already made a first, highly public "solved a Millennium Prize problem" claim (Navier–Stokes) on Sept. 8–9, 2026, but that problem is explicitly excluded by this market's rules, and as of early October 2026 it has not been independently verified , with the Clay Mathematics Institute still reviewing the proof — showing the process from "AI announcement" to any outside confirmation remains open-ended. https://en.wikipedia.org/wiki/Millennium_Prize_Problems ; https://www.kucoin.com/news/flash/openai-reports-major-progress-on-second-millennium-prize-problem - [2026-09-10] Within 48 hours of the Navier–Stokes announcement, OpenAI itself stated to the press 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] Reporting (The Information, via Gizmodo/the-decoder) says OpenAI employees expect the Hodge Conjecture to be solved "relatively soon," but this is a single anonymously-sourced claim, not an OpenAI announcement: the company is now working on the Hodge conjecture... Employees expect a solution soon, according to a person familiar with the matter. Any announcement could be delayed, though . https://the-decoder.com/openai-reportedly-closes-in-on-solving-the-hodge-conjecture-its-second-millennium-prize-problem/ - [2026-09-17] Analysts note OpenAI is deliberately slowing its rollout after backlash: After the PR crisis around Navier-Stokes, OpenAI wants to get the messaging right this time , suggesting institutional caution could delay or soften any second claim rather than accelerate it. https://the-decoder.com/openai-reportedly-closes-in-on-solving-the-hodge-conjecture-its-second-millennium-prize-problem/ - [2026-09-22] A business-analysis outlet summarized the evidentiary chain for the rumored second solve as multiple hearsay layers: The Information reports, sourced anonymously, that OpenAI is close to solving the Hodge Conjecture, with people familiar and staff expecting a solve "relatively soon." ... An X post named Hodge. The Information reported the Hodge attribution, anonymously sourced. Each increment of specificity was added by a different party with a different standard of evidence , underscoring that no confirmed second claim exists yet. https://fourweekmba.com/ai-openai-hodge-conjecture-clay-scoreboard/ - [2026-09-22] OpenAI separately announced a large batch of math results (the "100+ problems" announcement) that reportedly includes a second Millennium Prize problem: OpenAI says a new internal model has solved more than 100 long-standing problems across most areas of mathematics, on top of the Navier-Stokes Millennium Problem... Among the solved problems is reportedly a second Millennium Prize Problem, the Hodge conjecture — though this is reported/"reportedly," not an explicit, confirmed OpenAI claim naming Hodge. https://the-decoder.com/openai-says-its-internal-model-solved-over-100-long-standing-math-problems-after-just-a-month-of-training/ - [2026-09-19] Rival Anthropic's models have only reached "progress," not a solve, on the Riemann Hypothesis, illustrating that even frontier AI labs racing on these problems have mostly produced partial advances rather than qualifying full solutions: No. The Riemann Hypothesis remains unproven. What an unreleased Anthropic research version of Claude did was push forward a specific numerical bound tied to the problem . https://www.livescience.com/technology/artificial-intelligence/this-is-what-happened-with-claude-and-me-ais-failed-attempt-to-crack-the-riemann-hypothesis-led-mathematician-to-a-breakthrough - [2026-09-10] OpenAI is also reportedly pointing its Navier-Stokes-era internal model directly at Riemann Hypothesis and P vs NP, per NYT-sourced reporting: OpenAI is using the internal model that produced its Navier–Stokes proof to attempt the Riemann Hypothesis and P vs NP... The company is now using the internal model behind its Navier-Stokes proof on the Riemann Hypothesis and P versus NP , broadening the set of plausible "second solve" candidates beyond Hodge. https://huggingnews.com/ai/openai-targets-riemann-hypothesis-and-p-vs-np-with-navier-stokes-model-ff5bf602 - [2026-09-?/historical] Historical "solved" claims for these specific problems by non-AI mathematicians have a poor track record of holding up: e.g., De Branges subsequently circulated further manuscripts claiming proofs or extensions of the Riemann hypothesis; these claims have not been accepted. In June 2004, de Branges announced he had a proof of the Riemann hypothesis — a two-decade-old unresolved claim, showing that "announced solved" and "actually accepted" can diverge for years or indefinitely. https://en.wikipedia.org/wiki/Louis_de_Branges_de_Bourcia - [2026-earlier] Pre-2026 Metaculus/Manifold forecasts (from before the current AI math race accelerated) put "AI solves a Millennium Prize Problem" at a median year in the 2040s–2050s for most problems, e.g. P versus NP Jan 2068, Yang–Mills existence and mass gap Oct 2055, Birch and Swinnerton-Dyer conjecture Oct 2049, Riemann hypothesis Jun 2046 — these forecasts are now badly outdated given 2026 events, but illustrate how fast the outside view has shifted. https://www.metaculus.com/questions/15838/millennium-prize-problems-solution-date/ - [prior, ~2024] A related Manifold market, "OpenAI Claims to Solve another Millennium Prize Problem before 2027?", stood at roughly OpenAI Claims to Solve another Millennium Prize Problem before 2027? 72% chance historically (note: this snapshot is from well before the September 2026 events and is now stale, but shows forecasters already expected OpenAI-specific claims, not just any-AI claims, to be plausible within this timeframe). https://manifold.markets/DanielMittelman/will-an-ai-solve-a-millennium-probl BOTTOM LINE: The base rate for any entity producing a Clay-recognized "solved" Millennium Prize announcement is extremely low historically (one confirmed case in 26 years), but OpenAI itself has broken from that pattern by already issuing one high-profile "solved" claim (Navier–Stokes, excluded here) within days of a rumor, and has since publicly stated "substantial progress" on an unnamed second problem (widely rumored to be Hodge) with staff reportedly expecting a solve "relatively soon," albeit tempered by deliberate caution after PR backlash — this is an unusually active and fast-moving process compared to any prior era, making another OpenAI announcement plausible within the ~15-month window, but far from guaranteed given OpenAI's stated desire to avoid another credibility fight and the historical tendency of informal "proof announced" claims to stall for years before acceptance (or never be accepted). [… truncated]

YES means: Between now (2026-10-06) and 2027-12-31, OpenAI or an official representative issues a formal statement, backed by its researchers or models (alone or with outside researchers), explicitly declaring that one of the five qualifying problems (Riemann, P vs NP, Yang-Mills, Hodge, BSD) has been "solved" — not merely progressed on — and does not merely credit outside researchers. - [2026-09-08] OpenAI already demonstrated willingness to make a maximalist "solved a Millennium Prize problem" public announcement with a full paper and Lean formalization within days of model output, establishing the exact announcement mechanism (blog post + paper + formalization) that would satisfy this market's resolution criteria if repeated for a qualifying problem. https://openai.com/index/navier-stokes-solution/ - [2026-09-11] OpenAI's own on-the-record statement to the New York Times that it had made "substantial progress" on a second Millennium Prize Problem is, per a detailed claim-chain analysis, "the narrowest, most authoritative claim" and originates from the company itself, not from rumor — showing OpenAI is comfortable publicly confirming active work on a second qualifying problem. https://decrypt.co/378551/openai-progress-second-millennium-prize-math-problem - [2026-09-21] OpenAI's own blog post (not press rumor) confirms that a single internal model trained from August 28 to September 21 — just 24 days — "has now resolved more than 100 long-standing open problems across most areas of mathematics" on top of Navier-Stokes, with its own mathematicians "surprised" by the pace, demonstrating an internally-claimed capability growth curve steep enough that a second qualifying solve is plausible within months rather than years. https://techcrunch.com/2026/09/21/openai-forms-math-advisory-group-as-its-ai-resolves-more-than-100-open-problems/ - [2026-09-17] The Information's sourcing (via a direct OpenAI employee, not a random leaker) stated employees "expect to soon crack the Hodge Conjecture," indicating the rumored timeline for the next candidate announcement is weeks, not years, which fits comfortably inside the market's 15-month window. https://gizmodo.com/openai-reportedly-trying-to-solve-hodge-conjecture-amid-feud-with-math-community-2000813658 - [2026-09-21] OpenAI's creation of the nine-member Advisory Group on Mathematics and Artificial Intelligence — including Fields-tier figures Edward Witten, Timothy Gowers, and Martin Hairer — is explicitly tasked with "assessing the significance of new results and coordinating their release," i.e., OpenAI has now built the institutional machinery specifically to vet and publicize exactly this kind of announcement, rather than abandoning the math-publicity strategy after the Navier-Stokes backlash. https://techcrunch.com/2026/09/21/openai-forms-math-advisory-group-as-its-ai-resolves-more-than-100-open-problems/ - [2026-09-22] Reporting notes OpenAI's own admitted competitive motive — it says it "only took on the Millennium Prize Problem after hearing rumors that another team, partly made up of Anthropic researchers, had already cracked it" — indicating a race dynamic with Anthropic that creates pressure to publish a second claim quickly rather than sit on results indefinitely. https://the-decoder.com/openai-says-its-internal-model-solved-over-100-long-standing-math-problems-after-just-a-month-of-training/ - [2026-09-10] NYT-sourced reporting also places the Navier-Stokes-era internal model directly onto the Riemann Hypothesis and P vs NP, meaning there are at least three plausible candidate problems (Hodge, Riemann, P vs NP) any one of which reaching an OpenAI "solved" announcement would trigger YES, widening the path beyond a single bottleneck. https://huggingnews.com/ai/openai-targets-riemann-hypothesis-and-p-vs-np-with-navier-stokes-model-ff5bf602 - [2026-10-04] Even skeptical fact-check coverage, while stressing nothing is confirmed, concedes the baseline fact that OpenAI itself "made substantial progress on an unnamed Millennium problem" remains live and unretracted seven weeks after the claim was first made, i.e., OpenAI has not walked back or abandoned the second-problem effort. https://kingy.ai/news/openai-next-millennium-prize-problem-hodge/ BOTTOM LINE: OpenAI has already shown it will issue "solved" announcements for Millennium-class problems on a timescale of days to weeks, has an on-the-record internal claim of active, rapid progress on a second qualifying problem (widely reported as Hodge), has built formal infrastructure (the IAS advisory group) specifically to vet and release such announcements, and faces competitive pressure from Anthropic — collectively giving a concrete, evidence-backed path to a second, qualifying announcement well within the window to December 31, 2027. NOT FOUND: No OpenAI statement naming any qualifying problem as solved as of the latest search (early October 2026); no confirmed date for a prospective Hodge/Riemann/P-vs-NP announcement; no evidence of Yang-Mills work specifically.

YES means: OpenAI (or an official representative) formally announces, in its own voice, that it has solved one of the five qualifying problems (Riemann, P vs NP, Yang-Mills, Hodge, BSD) — not Navier-Stokes — before 2027-12-31, 11:59 PM ET. - [2026-10-04] The IEEE Spectrum "reckoning" piece, the newest dated source found, still frames any second solve only as rumor: OpenAI "is reportedly focusing on solving the Hodge conjecture," with no confirmed result. https://spectrum.ieee.org/millennium-prize-ai - [2026-09-21/22] OpenAI's own institutional response to the Navier-Stokes backlash was to create the Advisory Group on Mathematics and Artificial Intelligence (AGMAI) at Princeton's IAS — nine unpaid mathematicians whose explicit job is to "assess the significance of new results and coordinate how and when those results are released," which structurally inserts a new review/delay layer between any internal "solve" and a public OpenAI announcement. https://www.promptea.me/en/blog/openai-math-advisory-group-ias-no-pacing-authority - [2026-09-22] The same source confirms OpenAI explicitly carved pacing out of the group's mandate — "the group will not be responsible for advising us on how to pace our internal progress on mathematics" — meaning the company retains discretion to delay *disclosure* even if internal progress continues, consistent with reports that OpenAI is deliberately slow-walking the next announcement after the Navier-Stokes PR fallout. https://www.promptea.me/en/blog/openai-math-advisory-group-ias-no-pacing-authority - [2026-09-14] An early X rumor predicted an OpenAI Hodge announcement "this week or next" (mid-September); that specific predicted window has now passed by three weeks with no OpenAI statement, illustrating a pattern of rumor timelines repeatedly slipping. https://x.com/imjustnewatai/status/2099632027184017633 - [2026-09-21] Separately, the Clay Mathematics Institute's own written rules require that a proposed solution be published in a qualifying outlet and then wait at least two years, and achieve "general acceptance in the global mathematics community," before CMI will even consider it — meaning even a future OpenAI "Hodge solved" announcement would not convert into any institute-level validation within this market's window, though the market's resolution criteria only requires OpenAI's own announcement, not CMI confirmation, so this cuts only weakly against YES. https://www.claymath.org/millennium-problems/rules/ - [2026-10-03] A fact-check/explainer piece notes that even OpenAI's sweeping "100+ problems solved" claim (which reportedly includes Hodge) was released "without a public problem list, proofs, or defined criteria for what counts as 'resolved,'" and that outside verification of specific results is expected only "by early 2027," not immediately — suggesting any named, resolution-qualifying claim (distinct from the vague aggregate claim) may not crystallize into a clean, attributable announcement soon. https://tech-insider.org/openai-100-math-problems-solved-24-days-2026/ - [2026-09-22] Reporting on the AGMAI launch notes it drew "skepticism from observers" and that the panel "does not have the authority to influence the company's internal research pace, strategy, or development timeline" — meaning the governance fix does not force faster or more definitive public claims, but observers also stress it can slow careful, defensible releases rather than rushed ones, cutting toward caution before any new formal "solved" statement. https://www.whalesbook.com/news/English/technology/OpenAI-Forms-Math-Advisory-Board-Amid-Academic-Backlash/6ab192eb5aacb956d07be57d - [2026-09-17] Even the most specific reporting on a "second solve" concedes the chain is multiple layers of hearsay — "The Information reports, sourced anonymously, that OpenAI is close to solving the Hodge Conjecture... Each increment of specificity was added by a different party with a different standard of evidence" — with no OpenAI statement naming Hodge as of the latest findable reporting. https://fourweekmba.com/ai-openai-hodge-conjecture-clay-scoreboard/ - [historical] The only Millennium Problem ever formally recognized as solved (Poincaré, by Perelman) took roughly seven years from first proof-sketch to CMI's prize announcement, and current CMI rules still mandate a minimum two-year waiting period after qualifying publication plus community acceptance — underscoring how rare and slow-moving formal "solved" status is in this domain even when an underlying proof is credible, independent of OpenAI's PR choices. https://www.sciencenews.org/article/million-dollar-math-prize-awarded-not-necessarily-accepted BOTTOM LINE: As of October 6, 2026, every concrete fact points to OpenAI having made only a vague, unnamed "substantial progress" claim on a second Millennium Problem plus an unverified aggregate "100+ solved" claim, while its own new Princeton advisory body and PR-crisis caution create institutional friction that could delay rather than accelerate any clean, named, OpenAI-attributed "solved" announcement; no such formal announcement exists yet, and all specific Hodge/BSD "solved" claims remain anonymously-sourced rumor, not an OpenAI statement. NOT FOUND: No OpenAI blog post, paper, or press statement naming Hodge, Riemann, P vs NP, Yang-Mills, or BSD as solved; no public roster or output from the AGMAI group addressing a second Millennium claim; no news after October 4, 2026 updating the status.