As a leader in artificial intelligence, OpenAI's models maintain competitive performance across benchmarks in 2025, though their position faces growing challenges from global competitors. Released in April 2025, the o3 and o4-mini models are OpenAI's most capable AI models to date, designed to handle a broad range of tasks including coding, math, science, and visual perception with advanced reasoning capabilities and full tool access (OpenAI, Axios, TechTarget). These models have achieved top scores on some of the benchmarks used by the AI Index report including 93.4% on American Invitational Mathematics Examination 2024 and 20.3% on Humanity’s Last Exam (Vellum). However, OpenAI and U.S. firms face pressure from other countries internationally, as the gap between the top U.S. and Chinese AI models has shrunk significantly over the past year, largely thanks to China’s DeepSeek R1 (Visual Capitalist).