Decision models · Meta

Llama Guard 4 12B

Name: Llama Guard 4 12B
Author: Meta

Task: Safety and policy checks
Returns: Fixed labels
Precision tier: 66 / 100
Size: 12B
Max input: 128K tokens
License: llama4
Released: 2025-04

What it decides

Returns: safe or unsafe, plus the MLCommons hazard codes (S1 to S14) that apply; checks text and images.
Checks: Prompts, replies and images
Languages: 8 languages
Precision: 66 / 100 · F1 0.74 on WildGuardTest prompts and 0.70 on BeaverTails replies (as reported on the Shieldstral card).
Clarity: Fixed labels (12 of 15 clarity points)
How it runs: Writes a short answer · a small language model, so it runs in the same apps as LLMs.
License: llama4 (Custom (read the terms))

Source: model card on Hugging Face (opens in a new tab) · How decision models are rated

How to run it

Original weights · Hugging Face

Full-precision files from Meta, for fine-tuning, vLLM, Transformers.

huggingface.co/meta-llama/Llama-Guard-4-12B (opens in a new tab)

Free Hugging Face account needed: request access on the model page, and Meta approves it (usually within minutes to a few hours).

Links checked 3 days ago against Hugging Face, Ollama and GitHub.

Apps that run Llama Guard 4 12B

Our pick for each kind of computer, assuming the model fits in its memory. Pick your own GPU or Mac in the Model Finder for an exact answer.

Mac (Apple silicon)

Sentence Transformers

A few lines of Python with Transformers or Sentence Transformers; easiest for scripts and notebooks.

Also good: Docker Model Runner

Windows PC, NVIDIA GPU

Sentence Transformers

A few lines of Python with Transformers or Sentence Transformers; easiest for scripts and notebooks.

Also good: Docker Model Runner, vLLM

Windows PC, AMD Radeon

Sentence Transformers

A few lines of Python with Transformers or Sentence Transformers; easiest for scripts and notebooks.

Also good: Docker Model Runner

Linux, NVIDIA GPU

Docker Model Runner

Handy if you already work in Docker.

Also good: vLLM, Sentence Transformers

Linux server, 80 GB+ GPUs

vLLM

The full model fits in GPU memory at 8-bit, so it can batch many requests: best for serving a team or an app.

Also good: SGLang, Docker Model Runner

App	Type	Opens it as	Systems	Engine	Start
vLLM	Server	Original weights	WSLWindowsLinux	vLLM	`vllm serve meta-llama/Llama-Guard-4-12B`
SGLang	Server	Original weights	WSLWindowsLinux	SGLang	`python3 -m sglang.launch_server --model-path meta-llama/Llama-Guard-4-12B`
Docker Model Runner	CLI + API	Original weights	macOSWindowsLinux	llama.cpp, vLLM	`docker model run hf.co/meta-llama/Llama-Guard-4-12B`
Sentence Transformers	Library	Original weights	macOSWindowsLinux	PyTorch	Get Sentence Transformers

Which apps can open a model depends on the files published for it: GGUF for llama.cpp-based apps, MLX for Apple’s engine, the original weights for vLLM and SGLang. Compare every app.

Memory and speed

Each quantization on an NVIDIA GeForce RTX 4090, for one decision on a 512-token input. The best hardware table below covers other GPUs and Macs.

Format	Memory needed	Time per decision	Decisions per second	Quality	Download
`Q4_K_M`	8.3 GB	165 ms	6.0/s	62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.	All files (opens in a new tab)
`Q5_K_M`	9.5 GB	178 ms	5.6/s	64Good quality (B) · 64 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking.	All files (opens in a new tab)
`Q6_K`	10.8 GB	191 ms	5.2/s	65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking.	All files (opens in a new tab)
`Q8_0`	13.7 GB	222 ms	4.5/s	66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.	All files (opens in a new tab)
`F16`	25.0 GB	684 ms	1.5/s	66Good quality (B) · 66 of 100: how capable the model is at F16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking.	Original weights (opens in a new tab)

One request at a time. Servers that batch (Text Embeddings Inference, vLLM, Infinity) handle many more decisions per second.

Best hardware for Llama Guard 4 12B

The least expensive GPU in our catalog that runs it without offloading is the Intel Arc B570 (launch price $219) at Q4_K_M.

Quality	Hardware	Quant	Memory	Time per decision	Runs
65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA RTX PRO 6000 Blackwell	`Q6_K`	15.6 GB / 96.0 GB	119 ms	Runs great
65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 5090	`Q6_K`	12.4 GB / 32.0 GB	133 ms	Runs great
65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking.	AMD Radeon RX 7900 XTX	`Q6_K`	12.0 GB / 24.0 GB	187 ms	Runs well
65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 4090	`Q6_K`	12.0 GB / 24.0 GB	191 ms	Runs well
65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 5080	`Q6_K`	11.6 GB / 16.0 GB	244 ms	Runs well
65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking.	AMD Radeon RX 9070 XT	`Q6_K`	11.6 GB / 16.0 GB	259 ms	Runs well
65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 4080 SUPER	`Q6_K`	11.6 GB / 16.0 GB	282 ms	Runs well
65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 5070 Ti	`Q6_K`	11.6 GB / 16.0 GB	293 ms	Runs well
65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 4070 Ti SUPER	`Q6_K`	11.6 GB / 16.0 GB	323 ms	Runs well
65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 3090	`Q6_K`	12.0 GB / 24.0 GB	334 ms	Runs well

Check it on your own hardware

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.