Llama Guard 3 8B (Groq)
Llama Guard 3 8B is Meta's purpose-built safety classification model, fine-tuned to detect harmful content categories in both user inputs and model outputs. Running on Groq's LPU hardware, it delivers moderation decisions at speeds suitable for inline, real-time safety filtering rather than asynchronous batch review.
Unlike general-purpose models used secondarily for moderation, Llama Guard is trained specifically on safety taxonomies covering categories such as violence, hate speech, and self-harm content. Its 8B size strikes a balance between accuracy and inference speed, making it a practical guard layer that can sit in front of or alongside production LLM deployments.
Key Features
Classifies both user prompts and model responses for harmful content
Covers multiple safety categories including violence, hate, and self-harm
Real-time inference speed on Groq LPU for inline moderation
Purpose-trained safety taxonomy rather than general-purpose repurposing
Returns structured category labels for downstream filtering logic
Ideal Use Cases
Inline safety filtering layer for production LLM APIs
Automated content moderation in user-generated-content platforms
Compliance gatekeeping for regulated industries deploying AI
Red-teaming pipelines that need fast harm-category labeling
Dual-layer review of both input and output in chat systems
Example Prompts for Llama Guard 3 8B (Groq)
Technical Specifications
| Provider | Groq |
| Category | Text |
| Modality | Text -> Text |
Frequently Asked Questions
Try Llama Guard 3 8B (Groq) now
Start using Llama Guard 3 8B (Groq) instantly — 100 free credits, no credit card required. Access 750+ AI models through one platform.