Lost in the Middle: How Language Models Use Long Contexts

Our summary

An empirical study showing that LLM accuracy depends strongly on where relevant information sits in the input: performance is highest when key facts are at the very start or end of the context and degrades markedly when they fall in the middle, producing a characteristic U-shaped curve.

Why it matters

Directly undermines the assumption that a bigger context window means reliable recall — central to long-document and RAG use.

Cited by these methods

Related findings (1)

• Lost in the middle: degraded recall for mid-context informationMedium

  Retrieval accuracy is highest for facts at the start and end of a long context and drops for facts in the middle.

Cite this