/

← All benchmarks

INCLUDE

multilingual · regional-knowledge↗ Quelle

57Evidence

not yet measured

What it measures

Regional verankertes Wissen in 44 Sprachen, aus lokalen Pruefungen.

What it does not measure

Uebersetzung, Sprachqualitaet.

How the score is composed

Currency50

Discrimination50

Headroom50

Contamination safety60

Gaming resistance100

Methodological rigor80

Reproducibility20

Details

Benchmaxxing risk: low
Difficulty: medium
Public test set: yes
Human-verified: yes
LLM-judged: no
Skala: 0–100 % correct (höher besser)

Hand-maintained — no automatic adapter yet.

Best measured models

No results ingested yet.