The new AI model HRM surpasses large language models

Singapore-based Sapient Intelligence has introduced a new AI model called HRM, which, despite having only 27 million parameters, demonstrates exceptional performance on complex reasoning tasks—surpassing many larger language models. Inspired by the workings of the human brain, HRM avoids potential pitfalls of traditional “chain-of-thought” approaches by employing implicit reasoning pathways. In benchmark tests such as “Extreme Sudoku” and “Hard Mazes,” HRM significantly outperformed mainstream chain-of-thought models like o3-mini-high and Claude 3.7 Sonnet. Researchers note that HRM offers remarkable advantages in both data efficiency and reasoning capabilities, suggesting it could have profound implications for real-world enterprise AI applications.