Field Watch — Benchmark

LongMemEval

Last updated 7 October 2026

Why it matters

The authors publish 500 questions across information extraction, multi-session reasoning, temporal reasoning, knowledge updates and abstention.

Primary sources

ActiveContinuity/memory benchmark

How entries are chosen, categorized, and re-checked: How the map works.

Cite this page

How to cite: Cite the specific page and section (e.g., “The Continuity Layer, LongMemEval”), not the site as a general source. Attribute to The Continuity Layer.

Suggested citation: The Continuity Layer. “LongMemEval.” https://thecontinuitylayer.com/field-watch/longmemeval (last updated 2026-10-06).

Continuity layer — this site’s working definition: A continuity layer is the part of an AI system that carries forward governed state across sessions, tools, models, and hosts: what is currently true, what changed, why it changed, by whose authority, and which evidence supports it. It sits above memory and retrieval, and below agent execution, turning stored history into usable, auditable continuity; and it can correct, supersede, revoke, expire, and forget what it carries, with evidence that it did.