OpenAI announces solving a tough problem in 88 hours, smart money predicts AI will achieve another breakthrough within the year.

According to PolyBeats monitoring, on the prediction market Polymarket, 1 smart money address invested $2.0k in "Will an AI announce a solution to another Millennium Prize Problem this year?" on "Yes," with an average entry probability of 51.2%; the current "Yes" probability is 64.5%.

Rumpelstiltskin invested $2.0k, and the best-correlated category for this market is Artificial Intelligence, with a net category profit of $23.7k. In this category, their win rate across 182 settled trades is 107/182 (59%), including 30 trades with a buy price below $0.8 and a sell price above $0.95. Within a similar cost-price range ($0.451-$0.6), their historical median investment amount is $240.

The Millennium Prize Problems were established by the Clay Mathematics Institute in 2000, with a $1m prize for each problem. This market requires a qualified AI lab to announce within the year that it has solved the Riemann Hypothesis, P vs. NP, Yang-Mills existence and mass gap, the Hodge conjecture, or the Birch and Swinnerton-Dyer conjecture. Partial results and research progress do not count, and the Navier-Stokes equations are explicitly excluded by the rules. A lab only needs to make a claim of solving it.

This account's main profits come from Claude 5 release timing at about $6.9k, Grok 5 release timing at about $3.9k, Astra specific release date at about $3.6k, and model leaderboard markets at about $3.6k; the largest losses come from Best AI Model in February at about $10.1k and GPT-5.3 release timing at about $4.1k. The account currently buys "Yes" on AI labs and OpenAI announcing a solution within the year, while predicting that Anthropic will not announce a solution across multiple deadlines, with its judgment more concentrated on OpenAI.

Quanta Magazine reported on the 8th that OpenAI used about ten thousand autonomous AI agents to study the Navier-Stokes equations; after running for 88 hours, the agents produced a singularity proof, and another set of models then spent 17 hours formalizing the result into a Lean proof. The agents exchanged nearly five million messages in total, and the compute cost reached millions of dollars. The result still requires researchers to confirm that the formalized proposition is fully consistent with the original mathematical problem, and there is also controversy over attribution of the research contribution.