This market predicts whether Google DeepMind will announce an AI model that outperforms the average human (60% score) on the ARC-AGI-2 benchmark before January 15, 2026.
Context: Google DeepMind CEO Demis Hassabis stated in April 2025 that AGI could arrive within 5-10 years. The ARC-AGI-2 benchmark is considered one of the most challenging tests of general intelligence, with current AI models scoring far below human performance:
Average human score: 60%
Current top AI models: 1-3%
Resolution Criteria:
This market resolves to YES if Google DeepMind publicly announces an AI model that achieves a verified score higher than 60% on the ARC-AGI-2 benchmark before January 15, 2026, as confirmed on the official ARC Prize leaderboard (https://arcprize.org/leaderboard) or in an official announcement from Google DeepMind.
This market resolves to NO if by January 15, 2026, Google DeepMind has not announced any AI model scoring above 60% on the ARC-AGI-2 benchmark.
The model must be primarily developed by Google DeepMind (not a collaboration where another organization is the lead developer).