Meta AI’s model recently participated in the Asian Physics Olympiad’s theoretical exam, achieving a perfect score of 30/30, which placed it among the top three student contestants. This accomplishment showcases Meta AI’s advanced reasoning and multimodal capabilities, which are essential for addressing the complexities of the exam’s theoretical content. The APhO committee’s decision to allow the model to compete reflects the ongoing trend in AI benchmarking, where developers evaluate models through academic competitions to assess their problem-solving skills.

Meta AI: Meta AI is the artificial intelligence research division of Meta Platforms focused on developing large language and multimodal models. The division submitted one of its models to the Asian Physics Olympiad theoretical exam to demonstrate advanced reasoning capabilities. The model achieved a perfect score and tied with the top student performers.
APhO committee: The APhO committee organizes the Asian Physics Olympiad, an annual international competition for high school students emphasizing theoretical physics problems. The committee approved Meta AI’s request to enter a model in the exam alongside human contestants. This participation served as a demonstration of the model’s performance in a formal academic setting.

AI Benchmarking: AI developers routinely test models on academic competitions to evaluate reasoning and problem-solving skills.
Multimodal Capabilities: Meta AI highlighted its model’s multimodal features as key to handling the exam’s theoretical content.