Will any AI achieve a score of 25% on ARC-AGI-3 by the end of 2026?
11
Ṁ1kṀ1.6kDec 31
80%
chance
1H
6H
1D
1W
1M
ALL
Before betting, I recommend reading the scoring methodology page. It seems to be deliberately designed to produce lower scores: ARC-AGI-3 Scoring Methodology - ARC-AGI-3 Docs.
Scoring is based on the official leaderboard. As of market creation, the top score is around 0.3%:
This question is managed and resolved by Manifold.
Market context
Get 
1,000 to start trading!
1,000Sort by:
I'm assuming this is based on the official ARC-AGI leaderboard (not e.g. the Kaggle leaderboard)? Fyi for any other bettors, they aren't scoring harnesses. All official leaderboard scores are based on their official harness and prompts, so you can probably ignore any reports about a team you haven't heard of scoring super well on ARC-AGI-3. If this resolves YES it'll be because the frontier models get significantly better.
People are also trading
Related questions
Best ARC-AGI-3 score in 2026?
Which lab will be the first to reach 50% on ARC-AGI-3?
Above human scores on ARC-AGI-3 in 2026?
49% chance
Will an AI achieve >85% performance on the FrontierMath benchmark before 2028?
77% chance
Will any AI model score above 95% on ARC-AGI-2 by end of 2026?
72% chance
[ACX 2026] What will be the highest score achieved on ARC-AGI-2 before 2027?
93.9
When will the first model reach 50% on ARC-AGI-3?
9/14/28
Chatbot Arena: How high will AI score in 2026?
Will AI achieve AGI (Artificial General Intelligence) by 2030?
52% chance
Will an AI achieve a perfect score on the Miklós Schweitzer Competition before 2035?
81% chance