HRM-Symbolic
Open-sourced in July 2025, HRM-Symbolic is a reasoning model built for deep, efficient problem-solving through hierarchical latent-space reasoning.
Download HRM
HRM Reasoning in Action
Sudoku offers a simple, intuitive way to make reasoning visible. Here, it serves as a compact example of reasoning that HRM applies in high-impact domains.
Start with a random puzzle or enter your own.
Enter or paste your Sudoku puzzle below. Use “.” or “0” for empty cells.
Key Advantages
Greater Reasoning Depth
Multi-Scale, Multi-Step Reasoning
500-1000 effective layers solve complex problems through depth, not scale.
More Perceptive
Numerical & Pattern Sensitivity
Strong understanding of numbers and patterns for time-series and structured data.
Smarter with Less Data
Small-Sample Learning
Learns effectively from thousands of samples, not millions.
More Efficient
Adaptive Computation
ACT dynamically optimizes inference, reducing cost without sacrificing performance.
Smaller. Faster. Stronger.
Ultra Light. Superior Performance
0.027B parameters. No pretraining. No CoT. 100x faster reasoning. SOTA reasoning performance. Edge deployable.
Application Domains
Our architecture powers advanced reasoning across complex, high-impact real-world domains.
Benchmarks
Chain-of-thought, pretrained
Direct prediction, small-sample learning
ARC-AGI-1
960 training examples
ARC-AGI-2
1120 training examples
Sudoku-Extreme (9×9)
1000 training examples
Maze-Hard (30×30)
1000 training examples
Hierarchical Reasoning Model
Guan Wang
Jin Li
Yuhao Sun
Xing Chen
Changling Liu
Yue Wu
Meng Lu
Sen Song
Yasin Abbasi Yadkori