# The Tool-Overuse Illusion: Why Does LLM Prefer External Tools over Internal Knowledge?

arXiv: https://arxiv.org/abs/2604.19749 · HTML: https://mlsanity.com/paper/2604.19749/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 28 of its 35 references that are in the map.
- First posted on arXiv 3 Mar 2026.

## Builds on (most cited first, 28 of 28)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning](https://mlsanity.com/paper/1506.02142/) (2015)
- [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://mlsanity.com/paper/2503.14476/) (2025)
- [LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models](https://mlsanity.com/paper/2403.13372/) (2024)
- [Qwen2.5-Math Technical Report: Toward Mathematical Expert Model via Self-Improvement](https://mlsanity.com/paper/2409.12122/) (2024)
- [Program of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Tasks](https://mlsanity.com/paper/2211.12588/) (2022)
- [Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?](https://mlsanity.com/paper/2504.13837/) (2025)
- [API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMs](https://mlsanity.com/paper/2304.08244/) (2023)
- [ReTool: Reinforcement Learning for Strategic Tool Use in LLMs](https://mlsanity.com/paper/2504.11536/) (2025)
- [ToRA: A Tool-Integrated Reasoning Agent for Mathematical Problem Solving](https://mlsanity.com/paper/2309.17452/) (2023)
- [MetaTool Benchmark for Large Language Models: Deciding Whether to Use Tools and Which to Use](https://mlsanity.com/paper/2310.03128/) (2023)
- [MathCoder: Seamless Code Integration in LLMs for Enhanced Mathematical Reasoning](https://mlsanity.com/paper/2310.03731/) (2023)
- [SimpleTIR: End-to-End Reinforcement Learning for Multi-Turn Tool-Integrated Reasoning](https://mlsanity.com/paper/2509.02479/) (2025)
- [Buffer of Thoughts: Thought-Augmented Reasoning with Large Language Models](https://mlsanity.com/paper/2406.04271/) (2024)
- [Agent RL Scaling Law: Agent RL with Spontaneous Code Execution for Mathematical Problem Solving](https://mlsanity.com/paper/2505.07773/) (2025)
- [Unlocking the Capabilities of Thought: A Reasoning Boundary Framework to Quantify and Optimize Chain-of-Thought](https://mlsanity.com/paper/2410.05695/) (2024)
- [SMART: Self-Aware Agent for Tool Overuse Mitigation](https://mlsanity.com/paper/2502.11435/) (2025)
- [Understanding Tool-Integrated Reasoning](https://mlsanity.com/paper/2508.19201/) (2025)
- [Demystifying Reinforcement Learning in Agentic Reasoning](https://mlsanity.com/paper/2510.11701/) (2025)
- [Self-DC: When to Reason and When to Act? Self Divide-and-Conquer for Compositional Unknown Questions](https://mlsanity.com/paper/2402.13514/) (2024)
- [Towards Effective Code-Integrated Reasoning](https://mlsanity.com/paper/2505.24480/) (2025)
- [RL Grokking Recipe: How Does RL Unlock and Transfer New Algorithms in LLMs?](https://mlsanity.com/paper/2509.21016/) (2025)
- [Tool Zero: Training Tool-Augmented LLMs via Pure RL from Scratch](https://mlsanity.com/paper/2511.01934/) (2025)
- [WTU-EVAL: A Whether-or-Not Tool Usage Evaluation Benchmark for Large Language Models](https://mlsanity.com/paper/2407.12823/) (2024)
- [A2FM: An Adaptive Agent Foundation Model for Tool-Aware Hybrid Reasoning](https://mlsanity.com/paper/2510.12838/) (2025)

## Built on by (most cited first, 6 of 6)

- [When Top-K Misses the Decision: Tool-Call Drift in Multi-Teacher On-Policy Distillation](https://mlsanity.com/paper/2607.07050/) (2026)
- [From Knowing to Acting: Benchmarking Self-Awareness Capability of LLM Agents](https://mlsanity.com/paper/2606.20661/) (2026)
- [MemToC: Benchmarking Memory-Tool Conflict Resolution in Large Language Models](https://mlsanity.com/paper/2608.26295/) (2026)
- [Paying to Know: Micro-Transaction Markets for Verified Product Information in Agentic E-Commerce](https://mlsanity.com/paper/2606.24783/) (2026)
- [When Tools Get in the Way: The Effect of Unnecessary Tool Availability on LLM Answering](https://mlsanity.com/paper/2609.14157/) (2026)
- [OverAct: Measuring and Mitigating Proactive Over-Authorization in LLM Tool-Calling Agents](https://mlsanity.com/paper/2610.01508/) (2026)

## Abstract (arXiv)

Equipping LLMs with external tools effectively addresses internal reasoning limitations. However, it introduces a critical yet under-explored phenomenon: tool overuse, the unnecessary tool-use during reasoning. In this paper, we first reveal this phenomenon is pervasive across diverse LLMs. We then experimentally elucidate its underlying mechanisms through two key lenses: (1) First, by analyzing tool-use behavior across different internal knowledge availability regions, we identify a textitknowledge epistemic illusion: models misjudge internal knowledge boundaries and fail to accurately perceive their actual knowledge availability. To mitigate this, we propose a knowledge-aware epistemic boundary alignment strategy based on direct preference optimization, which reduces tool usage in by 82.8% while yielding an accuracy improvement. (2) Second, we establish a causal link between reward structures and tool-use behavior by visualizing the tool-augmented training process. It reveals that textitoutcome-only rewards inadvertently encourage tool overuse by rewarding only final correctness, regardless of tool efficiency. To verify this, we balance reward signals during training rather than relying on outcome-only rewards, cutting unnecessary tool calls by 66.7% (7B) and 60.7% (32B) without sacrificing accuracy. Finally, we provide theoretical justification in this two lenses to understand tool overuse.