---
id: "meta-llama/llama-guard-4-12b"
also: ["meta-llama/llama-guard-4-12b", "meta-llama/Llama-Guard-4-12B", "hf/meta-llama/Llama-Guard-4-12B"]
cheapest_list_usd_per_m: "0.144 in, 0.144 out, at deepinfra/bf16/flex on openrouter"
cheapest_effective_usd_per_m: "0.1436 in, 0.0 out, at deepinfra/bf16/flex on openrouter"
context: 163840
max_output: 16384
input: ["image", "text"]
output: ["text"]
released: "2025-04-30"
api: ["/api/v1/chat/completions", "/api/v1/completions", "/api/v1/messages", "/api/v1/responses"]
params: ["frequency_penalty", "logit_bias", "max_tokens", "min_p", "presence_penalty", "repetition_penalty", "seed", "stop", "temperature", "top_k", "top_p"]
endpoints: 6
gateways: "openrouter (4), hf (2)"
no_training: 4
trains: 0
retains_prompts: 0
---
# Meta: Llama Guard 4 12B

Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification. Similar to previous versions, it can be used to classify content in both LLM...