---
title: 'Lost in the Middle: How Language Models Use Long Contexts'
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
license_note: 'paper is arXiv non-exclusive distribution license (verified on arXiv abs page 2026-11-11); this pointer note is our own words and is CC BY 4.0; we link to the paper rather than reproduce it'
source: https://iinventedlight.com/library/notes/2307-03172-lost-in-the-middle.md
source_license: 'arXiv non-exclusive distribution license'
attribution: 'Pointer by iinventedlight (AI-assisted). Source: Nelson F. Liu, Kevin Lin, John Hewitt, Ashwin Paranjape, Michele Bevilacqua, Fabio Petroni, Percy Liang, Lost in the Middle: How Language Models Use Long Contexts (arXiv non-exclusive distribution license). This note links to the paper rather than reproducing it.'
author: 'iinventedlight (AI-assisted)'
date: 2026-11-11
status: published
approved_by: "Cole Withington, 2026-11-11 20:58 PT, Grok Bot chat widget t1717s7407: \"Publish all 100 including the LLMs Encode Clinical Knowledge note\""
checked_on: 2026-11-11
as_of: 2026-11-11
confidence: medium
untested: 'Pointer note only. Paper not reproduced. Abs page and license checked 2026-11-11.'
category: arxiv
topic_area: 'ai-research: memory-and-context'
tags: arxiv, long-context, memory, evals
arxiv_id: 2307.03172
alias: 2307.03172-lost-in-the-middle
type: pointer
format: pointer
sources:
  - id: S1
    title: 'Lost in the Middle: How Language Models Use Long Contexts'
    publisher: arXiv
    authors: 'Nelson F. Liu, Kevin Lin, John Hewitt, Ashwin Paranjape, Michele Bevilacqua, Fabio Petroni, Percy Liang'
    url: https://arxiv.org/abs/2307.03172
    version_read: 'v3 (20 Nov 2023)'
    accessed: 2026-11-11
    license: 'arXiv non-exclusive distribution license'
---
# Lost in the Middle: How Language Models Use Long Contexts

## Summary
[Lost in the Middle: How Language Models Use Long Contexts](https://arxiv.org/abs/2307.03172) (2023) documented a failure mode that still shows up in long-context systems: models often use the beginning and the end of a long prompt well, and miss evidence placed in the middle.

What it introduced: a controlled study of how position in the prompt affects the chance that a model will actually use a given piece of evidence, across several models and context lengths, with a clear lost-in-the-middle pattern.

What it led to: later long-context benchmarks and compaction research that treat usable context as shorter than the advertised window. Live notes: [ao-context-long-context-limits](https://iinventedlight.com/library/notes/ao-context-long-context-limits/), [ao-context-longbench-v2](https://iinventedlight.com/library/notes/ao-context-longbench-v2/), and [2404-06654-ruler](https://iinventedlight.com/library/notes/2404-06654-ruler/). Batch D also has a pointer to LongBench.

Where to read it: https://arxiv.org/abs/2307.03172 (v3, last revised 20 Nov 2023). Accepted in TACL, 2023.

The paper is posted on arXiv under the arXiv non-exclusive distribution license. This note links to it rather than reproducing it. The note itself is CC BY 4.0.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Pointer by iinventedlight (AI-assisted). Paper license: arXiv non-exclusive distribution; this note links to it and does not reproduce it.

---

Agents: [/agents.md](/agents.md) | [/.well-known/agent-library.json](/.well-known/agent-library.json).