Meta Model Scores Perfect 30/30 in APhO Theoretical Exam

Meta AI announced that a model they submitted participated in the theoretical exam of the Asian Physics Olympiad (APhO) and achieved a perfect score of 30/30, tying with the top three human students. This event drew attention because Meta explicitly framed it as a demonstration of the model's advanced reasoning and multimodal capabilities rather than a routine product launch, noting that the participation was permitted by the competition's organizing committee.

Key Details and Derivative Claims

According to Meta AI's official statement, the model completed the theoretical exam portion, resulting in a 30/30 score. In subsequent reposts, some authors provided further specifics: users @ThomasScialom and @himanshustwts identified the participating model as 'Muse Spark' and attributed the result to the '2026 APhO'. Furthermore, @himanshustwts described the achievement as winning a 'gold medal' or 'tying for first place', calling it an 'amazing turnaround' for Alexandr Wang's team. However, these specific details regarding the model name, year, and medal terminology were absent from the original official post summary.

Reactions

Authors who shared the news generally viewed it as strong proof of AI's reasoning capabilities in highly difficult scientific subjects. @ThomasScialom noted it as a step toward 'using AI systems to accelerate science', while @shuchaobi emphasized it as a display of Meta AI's breakthrough in advanced reasoning and multimodal capabilities.

2026-07-15 ~ 2026-07-15 · 6 related posts

• [source] Meta Model Scores Perfectly in Physics Olympiad — AIatMeta · 2026-07-15
• Unnamed Meta Model Wins Gold at APhO — himanshustwts · 2026-07-15

4 near-duplicate retellings: ThomasScialom · xiangyue96 · shuchaobi · baaadas