Meta · Moderation
Llama Guard 4 12B
Llama Guard 4 is a natively multimodal 12B safety classifier for classifying prompts and responses as safe or unsafe.
Evidence and sources
🤗 meta-llama/Llama-Guard-4-12BFair value (50.2). Costs per task about what its benchmark results predict. From 1 benchmark with measured task costs. Little data behind it, so treat it as a rough estimate.How Value for money worksReleased 23 Apr 2025
Training-data cutoff, as its sources state it
Maker HQ: US (according to Meta's own website)
All rankings:
#289 on OpenRouter this week
#291 on OpenRouter today
#310 on OpenRouter this monthSources: DeepInfra data, flex.ai API data, Models.dev data, OpenRouter data docs, FastRouter data, Hugging FacePrimary data from DeepInfra
from flex.ai API: hugging face id
from Models.dev: weights availability
from OpenRouter: knowledge cutoff
from FastRouter: description
from Hugging Face: release date
Input & output
as stated by the model's providersText
Images
Text
Also known as: deepinfra/meta-llama/llama-guard-4-12b, llama-guard-4, meta-llama/llama-guard-4-12b, meta/llama-guard-4-12b, nvidia_nim/meta/llama-guard-4-12b, replicate/meta/llama-guard-4-12b
Like this one. Tap or hover any dotted value, label or icon to see where it comes from and how it is measured. open details when tapped or hovered. How this data works