OpenAI's o3 AI model scores lower on a benchmark than the company initially implied | TechCrunchOpenAI's o3 model benchmark results are disputed, raising questions about transparency and testing practices.