Markets
Leaderboard
What will be the highest score achieved on ARC-AGI-2 before 2027? Between 93.1% and 96.0% (inclusive): real money 72%, play money 72% 96.1% or higher: real money 28%, play money 28% Less than 85.1%: real money 5%, play money 1% Between 85.1% and 89.0% (inclusive): real money 6%, play money 1% Between 89.1% and 93.0% (inclusive): real money 49%, play money 55% Volume: real money 229.68548676, play money 41867.54528144 Bettors: real money 14, play money 13 Categories: Finance, Artificial Intelligence Related markets China to try to take Taiwan using military force by the end of 2026? (ID: 133863) Volume: real 1711.95573343, play 248684.5794556 Bettors: real 35, play 510 Leading (real money): China to try to take Taiwan using military force by the end of 2026? 100% Leading (play money): China to try to take Taiwan using military force by the end of 2026? 8% Companies to go public in 2026 (ID: 233778) Volume: real 8594.60534041, play 127388.27313896 Bettors: real 74, play 81 Leading (real money): Anthropic 93% Leading (play money): SpaceX 94% What will be the price of Bitcoin at the end of 2026? (ID: 239444) Volume: real 4553.73222117, play 128423.53448695 Bettors: real 49, play 55 Leading (real money): Between $ 75,000 and $ 89,999.99 46% Leading (play money): Between $ 75,000 and $ 89,999.99 39%

What will be the highest score achieved on ARC-AGI-2 before 2027?

Market Rules

This event predicts the highest verified score achieved on the ARC-AGI-2 benchmark (Semi-Private Evaluation Set) by any AI system, as reported on the official ARC Prize leaderboard, at any time before January 1, 2027, 00:00:00 UTC. Each listed market below corresponds to one mutually exclusive percentage range of the maximum score; exactly one market will resolve "Yes".

The Primary Designated Source is the ARC Prize official leaderboard (arcprize.org/leaderboard), specifically the highest verified ARC-AGI-2 semi-private-set score listed at any time before January 1, 2027, 00:00:00 UTC. Only scores that ARC Prize publicly verifies (i.e., submitted, evaluated, and listed on the leaderboard) count. Self-reported, unverified, or test-set results that have not been validated by the ARC Prize do not count. Markets whose ranges have been exceeded by a qualifying score may resolve "No" before the deadline. If the leaderboard methodology changes substantially, Futuur might follow the same methodology to determine the value. Otherwise, this question will be annulled.

Editor's Note: The references to “private evaluation set” have been corrected to “Semi-Private Evaluation Set” to accurately identify the dataset used by the official ARC Prize leaderboard, which remains the designated resolution source. This clarification also makes explicit that the highest qualifying score listed at any time before the deadline counts.

Tags

People also viewed