Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Inference Latency on Multi-turn Adversarial Defense Latency Benchmark (inference)

4Latency (ms)

Llama-Prompt-Guard-2-86M

-8.5275.99160.5245.01Feb 18, 2026
Updated 4d ago

Evaluation Results

MethodLinks
2026.02
4
2026.02
4
2026.02
19
2026.02
43
2026.02
64
2026.02
77
2026.02
81
2026.02
125
2026.02
235
2026.02
317