OpenAI's o3 Model Achieves Human-Level Performance on Key Intelligence Test

The o3 system scored 88.5% on the ARC-AGI benchmark, raising questions about its implications for artificial general intelligence.