Prompt Repetition Improves Non-Reasoning LLMs

TL;DR

Repetition of prompts enhances the performance of non-reasoning large language models like Gemini, GPT, Claude, and Deepseek without additional computational cost.

Contribution

This paper demonstrates that prompt repetition can improve non-reasoning LLMs' performance, a simple technique previously underexplored.

Findings

Prompt repetition boosts accuracy for non-reasoning tasks.

Performance gains occur without increasing token count or latency.

Effective across multiple popular LLMs.

Abstract

When not using reasoning, repeating the input prompt improves performance for popular models (Gemini, GPT, Claude, and Deepseek) without increasing the number of generated tokens or latency.