ReasonIF: Large Reasoning Models Fail to Follow Instructions During Reasoning
By: Yongchan Kwon , Shang Zhu , Federico Bianchi and more
Potential Business Impact:
Makes AI follow instructions while thinking.
The ability of large language models (LLMs) to follow user instructions is central to their reliability, safety, and usefulness. While prior studies assess instruction adherence in the model's main responses, we argue that it is also critical for large reasoning models (LRMs) to follow user instructions throughout their reasoning process. Reasoning instruction following makes LRMs more controllable and transparent, while reducing risks of undesirable shortcuts, hallucinations, or reward hacking within reasoning traces. To evaluate this dimension, we introduce ReasonIF, a systematic benchmark for assessing reasoning instruction following. ReasonIF includes six categories of instruction prompts, spanning multilingual reasoning, formatting and length control. Across many open-source LRMs including GPT-OSS, Qwen3, and DeepSeek-R1, we find substantial failures in reasoning instruction adherence: the highest instruction following score (IFS) remains below 0.25, meaning that fewer than $25\%$ of reasoning traces comply with the given instructions. Notably, as task difficulty increases, reasoning instruction following degrades further. We also explore two strategies to enhance reasoning instruction fidelity. (1) multi-turn reasoning and (2) Reasoning Instruction Finetuning (RIF) using synthetic data. RIF improves the IFS of $GPT-OSS-20B$ from 0.11 to 0.27, indicating measurable progress but leaving ample room for improvement.
Similar Papers
When Thinking Fails: The Pitfalls of Reasoning for Instruction-Following in LLMs
Computation and Language
Makes AI follow instructions better by fixing reasoning.
Incentivizing Reasoning for Advanced Instruction-Following of Large Language Models
CV and Pattern Recognition
Teaches computers to follow tricky directions better.
Light-IF: Endowing LLMs with Generalizable Reasoning via Preview and Self-Checking for Complex Instruction Following
Computation and Language
Teaches computers to follow tricky instructions better.