Llama Guard 4 12B

Provider: Meta Llama US

Input price
$0.18
/ 1M tokens
Output price
$0.18
/ 1M tokens
Context
164K tokens
Max output 16K tokens

Overview

Llama Guard 4 12B is a specialized model from Meta Llama (US) targeting security, content moderation, or filtering workloads. It supports vision capabilities and JSON mode, offering a substantial context window and a generous maximum output length.

Strengths

  • Specialized capabilities implied by its 'Guard' designation for content filtering or safety tasks.
  • Supports vision for multimodal input understanding.
  • Offers JSON mode for structured output generation.
  • Very large 163840 token context window for extensive input processing.
  • Balanced input and output pricing at $0.18 per million tokens.
  • Generous 16384 token maximum output length.

Best use cases

  • Implementing content moderation pipelines for text and image data.
  • Automating safety checks on user inputs and model outputs.
  • Analyzing multimodal content for policy compliance with structured responses.
  • Building safety and filtering layers for custom large language model applications.
  • Detecting and flagging inappropriate content within long conversations or documents.

Pricing position

At $0.18/M for both input and output tokens, Llama Guard 4 12B is considerably more expensive than Meta Llama's general-purpose Llama 3.1 and 3.2 Instruct models. However, its input price is competitive with peers like Qwen3.6 Flash and GPT-5.4 Nano. Notably, its output price of $0.18/M is significantly lower than most similarly-priced peer models, which often charge $1.00/M or more for output. It offers pricing very similar to Ministral 3 14B 2512.

Caveats

While powerful for its niche, Llama Guard 4 12B lacks a free tier, increasing initial development costs. Its higher price point compared to other Llama 3/3.1/3.2 models makes it less suitable for general text generation or low-cost, high-volume tasks that do not require its specialized vision or safety capabilities. Consider more cost-optimized models for non-guard applications or when vision capabilities are not needed.

Notes drafted by an LLM from the figures on this page (2026-05-21) and rewritten when the price changes.

Key facts

Price basis
Median third-party host price (no first-party API)
Max output
16K tokens
Knowledge cutoff
2024-08-31
Listed on OpenRouter
2025-04-30
Tracked since
2026-05-08
Last updated
2026-11-10

Features

Image input

Monthly cost estimate

Scale Input (M) Output (M) Monthly
0.1M (hobby) 0.1 0.05 $0.03
1M (small app) 1 0.5 $0.27
10M (mid scale) 10 5 $2.70
100M (large) 100 50 $27.00

※ Cache and batch discounts not factored in. Detailed calculator

Official resources

Other models from Meta Llama

Similar-priced models

Source: OpenRouter API (meta-llama/llama-guard-4-12b)