Abstract
Long-horizon reasoning demands access to earlier information at a manageable generation cost. Full-history attention incurs growing storage and computation, while recurrent compression can lose precise details. Therefore, we present YANchor-4B, a general-purpose recurrent model that preserves crucial memory as ANchors for retrieval during subsequent reasoning. Beyond $O(N)$-time generation and $O(1)$ memory, YANchor enables effective long-horizon reasoning through its multidimensional memory mechanism. For example, on challenging math problems, it achieves 82.93% mean pass@1 on AIME 2024--2026 and 63.64% on HMMT, substantially outperforming linear-time, constant-state counterparts, including larger models. It also delivers several-fold higher batched long-generation throughput than Transformer and hybrid baselines on H100. Furthermore, evaluations across dozens of benchmarks demonstrate YANchor's superiority in general-purpose capabilities.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Ji, H., Xu, H., Zhang, W., & Ye, Q. (2026). YANchor-4B: Effective Long-Horizon Reasoning in O(N) Time with O(1) Memory. https://omanscience.com/en/articles/yanchor-4b-effective-long-horizon-reasoning-in-o-n-time-with-o-1-memory
MLA 9
Ji, Huishan, et al. "YANchor-4B: Effective Long-Horizon Reasoning in O(N) Time with O(1) Memory." https://omanscience.com/en/articles/yanchor-4b-effective-long-horizon-reasoning-in-o-n-time-with-o-1-memory.
Chicago (author–date)
Ji, Huishan, Hua Xu, Weiming Zhang, and Qirui Ye. 2026. "YANchor-4B: Effective Long-Horizon Reasoning in O(N) Time with O(1) Memory." https://omanscience.com/en/articles/yanchor-4b-effective-long-horizon-reasoning-in-o-n-time-with-o-1-memory.
Harvard
Ji, H., Xu, H., Zhang, W. and Ye, Q. (2026) 'YANchor-4B: Effective Long-Horizon Reasoning in O(N) Time with O(1) Memory', Available at: https://omanscience.com/en/articles/yanchor-4b-effective-long-horizon-reasoning-in-o-n-time-with-o-1-memory.
Vancouver
Ji H, Xu H, Zhang W, Ye Q. YANchor-4B: Effective Long-Horizon Reasoning in O(N) Time with O(1) Memory. https://omanscience.com/en/articles/yanchor-4b-effective-long-horizon-reasoning-in-o-n-time-with-o-1-memory
IEEE
H. Ji, H. Xu, W. Zhang, and Q. Ye, "YANchor-4B: Effective Long-Horizon Reasoning in O(N) Time with O(1) Memory," https://omanscience.com/en/articles/yanchor-4b-effective-long-horizon-reasoning-in-o-n-time-with-o-1-memory.