← All notes

Summary

LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding (2023) is a widely used bilingual, multi-task benchmark for long-context understanding. It gives a broader test than a single needle-in-a-haystack probe, across tasks and across English and Chinese.

What it introduced: a grouped suite of long-context tasks, covering retrieval, question answering, summarization, and other skills, designed so that models can be compared on more than one kind of long-input work.

What it led to: later long-context evals, including LongBench v2, and a habit of reporting "usable context" rather than only advertised window size. Live notes: ao-context-longbench-v2, ao-context-long-context-limits, and 2404-06654-ruler. See also the Lost in the Middle pointer in batch B.

Where to read it: https://arxiv.org/abs/2308.14508 (v2, last revised 19 Jun 2024). ACL 2024.

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.

@misc{iil_note_jx0890p0,
  author = {{'iinventedlight (AI-assisted)'}},
  title = {{LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding}},
  howpublished = {iinventedlight Library},
  year = {2026},
  date = {2026-11-11},
  url = {https://iinventedlight.com/library/notes/2308-14508-longbench/},
  note = {Permanent ID iil:note:jx0890p0. License: CC-BY-4.0.},
}

Markdown

Spotted a mistake? Use the MCP feedback tool or email hello@iinventedlight.com.

For language models and agents: start at /llms.txt | /agents.md | /.well-known/agent-library.json. This page in Markdown: /library/notes/2308-14508-longbench.md.