Trump announces 100% chip tariff as Apple ups U.S. investment
Investing.com -- Sapient Intelligence, comprised of former AI researchers from Google (NASDAQ:GOOGL), Anthropic, and more, has released an open-source AI model that achieves complex reasoning capabilities with minimal training data and parameters.
The company’s Hierarchical Reasoning Model (HRM) uses a brain-inspired architecture that leverages hierarchical structure and multi-timescale processing. Despite having only 27 million parameters and being trained on just 1,000 examples without pre-training, the model successfully tackles reasoning challenges that frustrate today’s large language models.
HRM incorporates three fundamental principles observed in cortical computation: hierarchical processing, temporal separation, and recurrent connectivity. It features a high-level module for abstract planning and a low-level module for detailed computations.
"AGI is really about giving machines human-level, and eventually beyond-human, intelligence," said Guan Wang, founder and CEO of Sapient Intelligence. "Our model actually thinks and reasons like a person, not just crunches probabilities to ace benchmarks."
In the Abstraction and Reasoning Corpus (ARC) AGI Challenge, HRM achieved 5% performance on ARC-AGI-2, outperforming larger models from OpenAI, DeepSeek, and Claude.
The company is exploring applications in healthcare for complex diagnostics, climate forecasting where it claims 97% accuracy in subseasonal-to-seasonal forecasting, and robotics as an on-device "decision brain."
Sapient Intelligence is headquartered in Singapore with research centers in San Francisco and Beijing. Its team includes former researchers from Google DeepMind, DeepSeek, Anthropic, and xAI, alongside academics from prestigious universities.
The source code is available on GitHub at https://github.com/sapientinc/HRM.
This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.