Lyraix-AI/LyraixGuard-v0
TEXT GENERATIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kPublished:Mar 28, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

Lyraix-AI/LyraixGuard-v0 is a 4 billion parameter Qwen3-4B based enterprise AI security classifier, fine-tuned to identify 13 types of security risks in user messages, including prompt injection and social engineering. This bilingual model (English and German) provides real-time classification as Safe, Unsafe, or Controversial, with optional reasoning traces. It is optimized for security gating in enterprise AI deployments, demonstrating high accuracy on external prompt injection benchmarks.

Loading preview...