Llama Prompt Guard 2 22M
22M-parameter (DeBERTa-xsmall) English classifier that detects prompt injection and jailbreak attempts
- Context window
- 512 tokens in · 512 out
The Llama Prompt Guard 2 line
2 releases in the Llama Prompt Guard 2 line.
BENCHMARKS
Not in the Epoch Capabilities Index. Epoch benchmarks what it can run, so this is a gap in coverage, not a poor result.
Meta's past year
8 release days in the past year, about one every 46 days. The next is expected around Dec 13, 2026. Forecasts ↗
NovDecJanFebMarAprMayJunJulAugSepOctNovDec
Specifications
RELEASE DATEApr 29, 2025
WEIGHTSOpen weights
CONTEXT WINDOW512
MAX OUTPUT512
REASONINGNot listed
TOOL CALLINGNot listed
INPUTStext
OUTPUTStext
Specifications and prices from Models.dev; availability and limits may differ by provider. Benchmarks from Epoch AI (CC BY 4.0), LiveBench and the other sources named above.