CoT-X: An Adaptive Framework for Cross-Model Chain-of-Thought Transfer and Optimization

TL;DR

This paper introduces CoT-X, an adaptive framework that compresses reasoning traces in large language models to enable efficient cross-model transfer and optimization, significantly reducing inference costs while maintaining high accuracy.

Contribution

The paper proposes a novel adaptive reasoning summarization framework that effectively compresses CoT traces for cross-model transfer, improving efficiency and robustness in resource-limited settings.

Findings

Up to 40% accuracy improvement over truncation at the same token budget.

Strong transferability across diverse model pairs and architectures.

84% reduction in evaluation cost through Bayesian optimization.

Abstract

Chain-of-Thought (CoT) reasoning enhances the problem-solving ability of large language models (LLMs) but leads to substantial inference overhead, limiting deployment in resource-constrained settings. This paper investigates efficient CoT transfer across models of different scales and architectures through an adaptive reasoning summarization framework. The proposed method compresses reasoning traces via semantic segmentation with importance scoring, budget-aware dynamic compression, and coherence reconstruction, preserving critical reasoning steps while significantly reducing token usage. Experiments on 7{,}501 medical examination questions across 10 specialties show up to 40% higher accuracy than truncation under the same token budgets. Evaluations on 64 model pairs from eight LLMs (1.5B-32B parameters, including DeepSeek-R1 and Qwen3) confirm strong cross-model transferability. Furthermore, a Gaussian Process-based Bayesian optimization module reduces evaluation cost by 84% and reveals a power-law relationship between model size and cross-domain robustness. These results demonstrate that reasoning summarization provides a practical path toward efficient CoT transfer, enabling advanced reasoning under tight computational constraints. Code will be released upon publication.