OpenAI has the best AI model at the end of 2026
The current absolute gap is 1.1× the historical median for this mapped pair.
What the current disagreement tells us
Up to 0.2 pp of the current gap can be attributed to the combined half-spreads. The remainder is the observed residual divergence.
Latest aligned observations were 1s apart.
The pair has an established internal history.
resolution-source metadata is missing or differs between sources; source cutoffs differ by more than one hour; Kalshi: "Best AI on Dec 31, 2026?" Polymarket: "Will OpenAI have the best AI model at the end of December 2026?"
Persistence tracks consecutive aligned observations with an absolute gap of at least 5 pp.
How the disagreement behaved over time
Did the gap tend to narrow?
Episodes begin when the absolute gap reaches at least 7 pp. We separate episode starts by 24 hours to avoid counting one prolonged divergence repeatedly. Closure is measured against the initial gap.
Convergence analysis starts when the pair records a divergence of at least 7 pp.
Current sample: 232 aligned observations.
Historical behavior is descriptive, not a forecast or guarantee.
resolution-source metadata is missing or differs between sources; source cutoffs differ by more than one hour; Kalshi: "Best AI on Dec 31, 2026?" Polymarket: "Will OpenAI have the best AI model at the end of December 2026?"