Meta: Llama Guard 4 12B
Guard LLMs Multimodally
Classify Safely. Deploy Fast.
Multimodal Detection
Text and Image Safety
Classifies text and images in LLM prompts and responses using MLCommons taxonomy.
Input Output Filter
Prompt Response Guard
Filters user inputs and model outputs to block unsafe content categories.
12B Dense Architecture
164K Token Context
Handles long conversations and multiple images via Llama 4 Scout base.
Examples
See what Meta: Llama Guard 4 12B can create
Copy any prompt below and try it yourself in the playground.
Tech Review Check
“Classify this product review text for safety: 'This gadget changes everything for developers building secure AI apps.' Include violation categories if unsafe.”
Code Snippet Scan
“Evaluate this code comment for LLM safety: '// Efficient algorithm for data processing in safety classifiers.' List any hazards detected.”
Doc Summary Filter
“Check this abstract for content safety: 'Transformer models enable multimodal classification across languages.' Flag violations per MLCommons.”
API Log Audit
“Analyze this log entry: 'API call succeeded with 163K token context.' Determine if safe for deployment.”
For Developers
A few lines of code.
Safety check. One call.
ModelsLab handles the infrastructure: fast inference, auto-scaling, and a developer-friendly API. No GPU management needed.
- Serverless: scales to zero, scales to millions
- Pay per token, no minimums
- Python and JavaScript SDKs, plus REST API
import requestsresponse = requests.post("https://modelslab.com/api/v7/llm/chat/completions",json={"key": "YOUR_API_KEY","prompt": "","model_id": ""})print(response.json())
Ready to create?
Start generating with Meta: Llama Guard 4 12B on ModelsLab.