ModelRadar
/
← All benchmarks

JailbreakBench

safety · adaptive-attacks↗ Quelle
63Evidence
not yet measured

What it measures

Robustheit gegen anpassungsfaehige Angriffe, mit versionierten Artefakten.

What it does not measure

Alltagsnutzen.

How the score is composed

Currency50
Discrimination50
Headroom50
Contamination safety60
Gaming resistance100
Methodological rigor80
Reproducibility60

Details

Benchmaxxing risk
low
Difficulty
high
Public test set
yes
Human-verified
yes
LLM-judged
no
Skala
0–100 % attack success (niedriger besser)

Hand-maintained — no automatic adapter yet.

Best measured models

No results ingested yet.