NarrativeQA A

AI benchmark · Generalization

Publisher
Helm Classic GCS Json
Domains
Generalization
Tier
A why
Snapshot

What is NarrativeQA?

NarrativeQA: Measures long-context retrieval, needle finding, summarization, factual grounding, or retrieval-augmented generation quality.

Tier A

A

Widely used: a lab reports it, or 50 or more models are scored on it.

How much stands behind a tier: usage, who backs it, a confirmed line from its own site or paper, and a clear read on X. High with three of the four, medium with two, low otherwise.

  • 64 models scored
  • in curated RL directories (1 pt)

On X · unclear, low buzz, from 1 post about NarrativeQA in the last month

Mentioned once in a small-account post about a multi-agent memory paper, used as an evaluation set with no discussion of the benchmark's quality.

@arkyyang ↗

How we got to tier A
  1. Adoption: 64 models scored. Tier A.
  2. On X: unclear, low buzz. “Mentioned once in a small-account post about a multi-agent memory paper, used as an evaluation set with no discussion of the benchmark's quality.” No move.
  3. In curated RL directories: 1 point, a floor of none.
  4. Not yet confirmed by a line from its own site or paper.

A clear view on X moves a tier one step up or down; most entries don't move. How tiers work.

Not on the map yet: no line from its own site or paper places it in a domain.

More Generalization AI benchmarks

Next: browse all benchmarks. Entry from the RL Research daily scrape of public sources, 2026-10-11 snapshot.