Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Prompt Repetition Improves Non-Reasoning LLMs

About

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.

Yaniv Leviathan, Matan Kalman, Yossi Matias• 2025

Related benchmarks

TaskDatasetResultRank
General KnowledgeMMLU (test)
Accuracy60.3
71
Science Question AnsweringARC Challenge (test)
Accuracy82.8
60
Question AnsweringSciQ (test)
Accuracy92.9
36
Long-context evaluationRULER 4k--
35
KnowledgeOpenBookQA (test)
Accuracy78
19
Complex ReasoningGSM8K (test)
Accuracy87.3
8
General knowledge and scientific retrievalMedQA (test)
Accuracy51
8
Complex ReasoningMMLU Pro (test)
Accuracy27.3
8
Long-context evaluationRULER 8K context length
Accuracy82.3
4
Long-context evaluationRULER 16k context length
Accuracy72.3
4
Showing 10 of 10 rows

Other info

Follow for update