OpenAI has introduced its latest artificial intelligence models, o3 and o3-mini, which are successors to the o1 reasoning model family. While these new models are not yet widely available, safety researchers can sign up for a preview. The o3 model is noted for its enhanced reasoning capabilities and aims to improve accuracy in complex tasks, although it is expected to come at a high cost. Additionally, the o1 model is now accessible on the Perplexity platform, marking a significant development in OpenAI's offerings. The o3 system has reportedly achieved human-level performance on a general intelligence test, raising discussions about the implications for artificial general intelligence (AGI) and its governance.