OpenAI’s o3 AI Model Falls Short of Benchmark Claims in FrontierMath Test

OpenAI’s o3 artificial intelligence (AI) model, which was released last week, is underperforming on a specific benchmark. Epoch AI, the company behind the FrontierMath benchmark, highlighted that the publicly available version of the o3 AI model scored 10 percent on the test, a much lower value than the company’s claim at launch.