Metraipolis › archive › dc_e8d58903bcce
Scaling of Capability and Efficiency at Inference Time in Large Reasoning Models — arXiv:2609.27166
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.27166
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T10:52:04.570Z
- SHA-256 of the archived text
83740cb909ea6ff16f6cbeba6338bc4e2512a7da022d84f93373c3b0d05368d2- Stable URL of this page
https://metraipolis.com/documento/dc_e8d58903bcce
How to cite
Scaling of Capability and Efficiency at Inference Time in Large Reasoning Models — arXiv:2609.27166 — arXiv (en), https://arxiv.org/abs/2609.27166 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e8d58903bcce (Metraipolis archive dc_e8d58903bcce, sha256 83740cb909ea6ff16f6cbeba6338bc4e2512a7da022d84f93373c3b0d05368d2)
BibTeX
@misc{dce8d58903bcce,
title = {Scaling of Capability and Efficiency at Inference Time in Large Reasoning Models — arXiv:2609.27166},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.27166},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e8d58903bcce; sha256 83740cb909ea6ff16f6cbeba6338bc4e2512a7da022d84f93373c3b0d05368d2},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.27166. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Capability and efficiency are two key dimensions of reasoning in large language models (LLMs). Capability refers to the ability to solve a given problem correctly, whereas efficiency refers to the ability to do so with limited resources. When LLMs use Chain-of-Thought (CoT) reasoning to solve problems of controlled hardness, both the number of problems solved correctly and the number of tokens required to reach a correct answer depend on problem hardness and model size. However, how these factors jointly shape capability and efficiency remains poorly understood. Here, we use hierarchical Bayesian models to evaluate the capability and efficiency of LLMs from the DeepSeek-R1-Distill model family across four classes of arithmetic and algorithmic reasoning problems. At a fixed model size, the probability of correctly solving an instance decays approximately exponentially with instance size, our proxy for problem hardness. The decay scale grows sublinearly with model size, indicating that larger models are more capable, but that capability gains diminish with scale. Output length grows as a power law with instance size, which serves as a proxy for difficulty. However, the parameters of this power law do not vary systematically with model size, suggesting that larger models do not become more efficient. Together, these findings reveal potential limitations of naive scaling as a strategy for developing more capable AI systems: capability improves with diminishing returns, while efficiency shows little to no improvement. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-24 in cs.LG). Categorie arXiv: Machine Learning (cs.LG) ; Physics and Society (physics.soc-ph). Pagina del lavoro: https://arxiv.org/abs/2609.27166 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-08T10:06:16.702Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).