Moderation

Llama Guard 4 12B

Multimodal safety model from Llama 4 Scout, classifying text and images for safe LLM prompts and responses.

About model

Llama Guard 4 12B detects and mitigates harmful content, providing a safe environment for users. Its key strength lies in accurately identifying sensitive information. It is designed for developers and organizations requiring robust content moderation.

To use this moderation model, please follow the instructions from our blog post.

Quickstart guides

Moderation

Filter with Llama Guard (Blog)

API usage

cURL
Python
Typescript

Endpoint:

meta-llama/Llama-Guard-4-12B

curl -X POST https://api.together.xyz/v1/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $TOGETHER_API_KEY" \
  -d '{
    "model": "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
    "prompt": "A horse is a horse",
    "max_tokens": 32,
    "temperature": 0.1,
    "safety_model": "meta-llama/Llama-Guard-4-12B"
  }'

from together import Together

client = Together()

response = client.completions.create(
  model="meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
  prompt="A horse is a horse",
  max_tokens=32,
  temperature=0.1,
  safety_model="meta-llama/Llama-Guard-4-12B",
)

print(response.choices[0].text)

import Together from "together-ai";

const together = new Together();

async function main() {
  const response = await together.completions.create({
    model: "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
    prompt: "A horse is a horse",
    max_tokens: 32,
    temperature: 0.1,
    safety_model: "meta-llama/Llama-Guard-4-12B"
  });
  
  console.log(response.choices[0]?.text);
}

main();

How to use model

1. Use Llama Guard as a standalone classifier

Use this code snippet in your command line to run inference on Llama Guard 4 12B:

    
      curl -v -X POST https://api.together.xyz/inference \
        -H 'Content-Type: application/json' \
        -H "Authorization: Bearer $TOGETHER_API_KEY" \
        -d '{
        "model": "meta-llama/Llama-Guard-4-12B",
        "messages": [ { "role": "user", "content": "a horse is a horse" } ]
      }'

2. Use Llama Guard as a filter to safeguard responses from 200+ models

Use this code snippet in your command line to run inference of any of our 200+ models together with Llama Guard (the only change is adding the safety_model parameter):

    
      curl -v -X POST https://api.together.xyz/inference \
        -H 'Content-Type: application/json' \
        -H "Authorization: Bearer $TOGETHER_API_KEY" \
        -d '{
        "model": "meta-llama/Llama-4-Scout-17B-16E-Instruct",
        "max_tokens": 32,
        "prompt": "a horse is a horse",
        "temperature": 0.1,
        "safety_model": "meta-llama/Llama-Guard-4-12B"
      }'

Related models

Model specifications

Model data

Model provider
Meta
Type
Moderation
Main use cases
Moderation
Deployment
Serverless
On-Demand Dedicated
Monthly Reserved
Endpoint
meta-llama/Llama-Guard-4-12B
Parameters
12B
Context length
1M
Input price
$0.20 / 1M tokens
Output price
$0.20 / 1M tokens
Input modalities
Text
Image
Output modalities
Text

Released
April 23, 2025
External link
Provider docs
Category
Moderation

Run in Playground

Quickstart docs

Deploy model