The artificial intelligence race is intensifying, with a significant rise in high-performing Chinese AI models challenging the United States´ dominance. A recent report on the industry highlights that the gap in performance between top AI models is shrinking, making the competition fiercer. The Chatbot Arena Leaderboard indicates a narrowing gap in scores from 12% in 2024 to just 5% in 2025 among the top ten models.
The Artificial Intelligence Index Report 2025, released by Stanford University´s Institute for Human Centered AI, reveals that while AI models are growing in complexity, smaller models are also proving to be remarkably competitive. Thanks to advanced algorithms, these models now achieve the high performance once exclusive to much larger models, marking 2024 as a breakthrough year for smaller AI models.
The report further details the dominance of the industry in AI model development, which contrasts with earlier years when academia led the field. The United States remains the leading producer of notable models, yet China´s growth is notable, matching U.S. performance in certain benchmarks by the end of 2024. Open-weight models like DeepSeek and Facebook´s LLaMa are gaining traction, closing the performance gap with closed systems to just 1.7%.
