Would you bet…
AI model scores ≥ 90% on FrontierMath Benchmark before 2027? Predictions
A YES share pays out if this happens and NO pays out if it doesn’t — so the 81% price is just the market’s implied chance of YES. How YES/NO contracts work →
- Platform
- Polymarket
- Volume
- $106,938 volume
- Resolves
- 31 Dec 2026
- Updated
- 6 days ago
The market prices a strongly favored outcome at 81%, suggesting a substantial consensus that some AI model will clear 90% on FrontierMath by end of 2026. That’s less than two years away. The price down 7 points over the past week, signaling modest skepticism creeping in—or simply profit-taking after a strong run.
What this hinges on: the gap between current SOTA performance and that 90% threshold. FrontierMath is a genuinely hard benchmark, designed to test reasoning on competition-level problems. We’d need to see either a major architectural leap or a model trained specifically for this task. The timeline is tight but not impossible given the pace of capability gains since 2023. EpochAI will be the scorekeeper.
81% at strongly favored levels reflects real progress in long-chain reasoning, but also real uncertainty about whether “hard math” remains a frontier or becomes routine by 2027. Volume of $107k is moderate, so any material new information—a model release, a public benchmark run—could move this meaningfully in either direction. The price is a live read, not a lock.
FAQ
What does a 81% price mean?
It is the market-implied probability. A 81% YES price means traders collectively judge the event about 81% likely.
How does this market resolve?
This market will resolve to "Yes" if a state-of-the-art (SOTA) AI model achieves a score of 90% or greater on the FrontierMath Exam by December 31, 2026, 11:59 PM ET. Otherwise, the market will resolve to "No". The primary resolution source will be information from EpochAI however a consensus of cr
Where can I trade it?
This market is listed on Polymarket. Prediction markets carry real financial risk and may not be available in every state.
What economic events can I trade?
Fed meetings, CPI and PCE inflation, nonfarm payrolls, unemployment, GDP and recession calls are the most liquid.
How is this different from futures?
Event contracts are simple binary yes/no positions priced from $0 to $1, rather than leveraged futures — easier to size and read as probabilities.
Which platform is best for economics?
Kalshi has the broadest macro slate; see our Kalshi review.
What is a prediction market?
A prediction market lets you trade contracts on whether a real-world event will happen. The live price moves with supply and demand and reads as the implied probability. Read more →
How do the odds work?
Every price between 1¢ and 99¢ is the implied chance of YES. A contract settles at $1 if it resolves yes and $0 if it does not. Read more →
Prediction market contracts carry real financial risk and can resolve to zero. 18+.
Before you trade
Read our independent reviews of the platforms behind these markets.