Security gateway
Content safety with Llama Guard
Llama Guard judges conversations against the S1 to S14 hazard categories, and you choose which to enforce.
What it is
Llama Guard 3 or 4 served on your own vLLM judges conversation turns against the S1 to S14 hazard categories, and you choose which categories to enforce.
Why you want it
Organizations with acceptable-use obligations need a record of harmful requests and a way to stop them, without a hosted moderation service.
How it works
- Full MLCommons hazard taxonomy with severity tiers
- Requests can be blocked by category
- Responses are judged in context of the prompt and logged without being withheld
See it on your own network.
The Community edition is free for up to 25 people. The 30-day Business trial unlocks every Business feature.