How the contract works
This contract settles at $1 if an eligible AI lab publicly claims, by 31 December 2026, a full solution to one of the five qualifying problems, with the lab's own researchers or models credited rather than outside mathematicians alone. It settles at nothing otherwise, including if only partial progress is announced or if the claim concerns the Navier-Stokes problem, which is explicitly excluded here. A price of 0.30, for example, would mean the market sees roughly three chances in ten that this happens before the deadline; that is a hypothetical, not this market's current level. Resolution follows Polymarket's rules, which point to the Clay Mathematics Institute's published problem list, and a position in this contract can typically be sold before settlement at whatever price the market shows at that time.
Yes19%
No81%
Probability
How the price has moved
The tracked venue currently prices this at 19% on $41,641 of volume, a modest figure that suggests limited participation rather than a thoroughly contested consensus. No detailed intraday or weekly price history was available for this market, and no single reported event, paper or announcement stands out as a trigger for the current level. The thinness of the volume is itself informative: it means the price mainly reflects how a small number of traders weigh the base rate of unsolved Millennium Problems against the pace of AI mathematical progress, rather than a broad market judgment.
Analysis
Context
The Clay Mathematics Institute named seven Millennium Prize Problems in 2000, each carrying a $1,000,000 award for a complete, verified solution. Only one, the Poincaré Conjecture, has been solved, by Grigori Perelman in 2003; he later declined both the prize and a Fields Medal. The other six, including the Riemann Hypothesis (open since 1859) and P versus NP (formalized in 1971), remain unsolved after decades of effort by professional mathematicians.
The consensus across the tracked venue sits at 19%, built on $41,641 of trading volume, a relatively thin market for a question this specific. That size matters: with one venue and modest volume, the price reflects a small number of participants weighing a narrow, binary technical claim rather than a broad consensus view, so it should be read as a rough signal rather than a settled estimate.
The case for a low number starts with base rates. In the roughly 26 years since the Clay Institute announced the seven problems, only one has been solved, and that was by a human mathematician working largely alone, not by an automated system or research lab. None of the five problems still eligible here (Riemann, P vs NP, Hodge, Birch and Swinnerton-Dyer, Yang-Mills) has produced even a widely credited near-solution, let alone one from an AI lab. Large language models and reasoning systems from OpenAI, Google DeepMind and Anthropic have shown improving performance on competition mathematics and formal proof benchmarks, but a full, peer-scrutinized proof of an open Millennium Problem is a different order of difficulty than solving bounded, previously-solved competition problems.
The resolution criteria also narrow the path to Yes. The rules require the lab's own researchers or models to be credited, not merely an outside academic citing AI assistance, and they require a claim of complete solution rather than partial progress, which is the more common outcome in mathematics even when real advances occur. Verification adds further friction: a serious Millennium Problem claim would need independent mathematical scrutiny before most observers, including market participants, would treat it as credible, and that process does not run on a calendar tied to a single year.
Working against the low price is the pace of investment in AI mathematical reasoning through 2025 and 2026, with labs explicitly targeting formal theorem-proving and research-level mathematics as a benchmark of progress. A single credible announcement, even if contested, could move this market quickly given its current thin volume, which is itself a reason the price can swing on sparse information.
What moves the probability
Base rate of Millennium solutions
Only one of the seven original Millennium Prize Problems has been solved since 2000, by a human mathematician, not a lab or model. That 25-year history of near-total non-resolution weighs heavily against any one-year window, pushing the probability down.
AI reasoning progress
Labs including OpenAI, Google DeepMind and Anthropic have pushed reasoning models toward harder formal mathematics and competition-style proofs. Continued, visible progress on that front is the main force that could pull this price up before the 2026 deadline.
Verification and crediting requirements
The rules require the lab's own researchers or models to be credited and demand a full solution, not partial progress. This raises the bar beyond a promising preprint, since mathematical claims of this scale normally take time for the community to check.
Thin market, low volume
With trading volume around $41,641 concentrated on a single venue, the price can move sharply on a single credible rumor or paper, since few participants are needed to shift it. That also means the current level reflects a small set of views rather than broad consensus.
Narrow annual deadline
The window closes 31 December 2026, so a solution announced even weeks later would not count for this contract, separate general-threshold market. That calendar cutoff compresses the odds relative to an open-ended question about whether AI ever solves one of these problems.
The case for
- A major lab would need to publicly claim a complete, credited solution to one of the five eligible problems before 31 December 2026, 11:59 PM ET.
- Investment in AI systems aimed at formal mathematics and theorem-proving has grown through 2025 and 2026, and a credible breakthrough on one narrow but tractable sub-problem could qualify under the rules.
- The resolution only needs the lab's own researchers or models to be credited, including jointly with outside collaborators, which is a lower bar than requiring the lab to work entirely alone.
- A single widely reported announcement, even if later disputed, could be enough to resolve the contract if it meets the Clay Institute's stated criteria at settlement.
The case against
- In roughly 26 years only one Millennium Prize Problem has been solved at all, and that was by an individual mathematician, not an automated system.
- Full proofs of open problems like the Riemann Hypothesis or P versus NP have resisted the entire mathematical community for over a century and over 50 years respectively.
- The rules exclude partial progress and exclude the Navier-Stokes problem entirely, narrowing the paths to a Yes outcome.
- Genuine mathematical claims of this magnitude typically require extended independent verification, which is difficult to complete and credit within a single calendar year.
What to watch
Watch for any formal preprint or lab announcement from OpenAI, Google DeepMind, Anthropic or a comparable AI research organization claiming a full solution to the Riemann Hypothesis, P versus NP, the Hodge Conjecture, the Birch and Swinnerton-Dyer Conjecture, or the Yang-Mills existence and mass gap problem. Also relevant are major AI mathematics benchmark results and conference papers through 2026, and any public statements from the Clay Mathematics Institute itself about submissions or claims under review. The hard cutoff is 31 December 2026, 11:59 PM ET, after which no further claims count for this contract.
Trade this contract
Open on Polymarket
- gas covered