Do LLMs Overthink Basic Math Reasoning? Benchmarking the Accuracy-Efficiency Tradeoff in Language Models

TL;DR

This paper introduces LLMThinkBench, a benchmark and empirical study analyzing the tradeoff between accuracy and verbosity in LLMs' basic math reasoning, revealing that longer reasoning does not always improve performance.

Contribution

The paper formalizes the accuracy-verbosity tradeoff, introduces the Overthinking Score, and provides a large-scale evaluation of 53 LLMs with open-source tools and benchmarks.

Findings

Model performance on complex benchmarks does not translate to basic math reasoning.

Reasoning models generate ~18x more tokens but sometimes have lower accuracy.

Extended reasoning budgets yield diminishing returns, with some models showing no accuracy gain.

Abstract

Large language models (LLMs) achieve impressive performance on complex mathematical benchmarks yet sometimes fail on basic math reasoning while generating unnecessarily verbose responses. In this paper, we present LLMThinkBench, a systematic benchmark and comprehensive empirical study to evaluate the efficiency of reasoning in LLMs, focusing on the fundamental tradeoff between accuracy and overthinking. First, we formalize the accuracy-verbosity tradeoff. Second, we introduce the Overthinking Score, a harmonic-mean metric combining accuracy and token-efficiency for holistic model evaluation. Third, we establish an evaluation protocol with dynamically-generated data across 14 basic math tasks. Fourth, we conduct a large-scale empirical study evaluating 53 LLMs, including reasoning and quantized variants across different reasoning budgets. Fifth, we release LLMThinkBench as an open-source Python package and public leaderboard for reproducibility. Our findings reveal: 1) model performance on complex benchmarks does not translate directly to basic math reasoning; 2) reasoning models generate ~18x more tokens while sometimes achieving lower accuracy and exhibit catastrophic collapse when tokens are constrained, dropping by up to ~36%; 3) the accuracy-verbosity relationship is non-monotonic with extended reasoning budgets yielding diminishing returns (GPT-5/o-series models show zero accuracy gain from low -> medium -> high reasoning effort). Our findings challenge the assumption that longer reasoning in LLMs necessarily improves mathematical reasoning. Our public leaderboard is available at https://ctrl-gaurav.github.io/LLMThinkBench/. Our open-source Python package is available at https://pypi.org/project/llmthinkbench/, and the codebase can be found at https://github.com/ctrl-gaurav/LLMThinkBench for easy and reproducible evaluation.