The deciding difference is operational: Rebuff can run on infrastructure you control, Lakera Guard cannot. If that constraint is real for you, it settles the choice before anything else is considered.
Prompt-injection and jailbreak detection as a low-latency API call.
Worth it the moment your agent reads text written by someone who is not your user.
Open-source prompt-injection detector with canary tokens.
| Lakera Guard | Rebuff | |
|---|---|---|
| Licence | Proprietary | Open source |
| Pricing | freemium | open source |
| Self-hostable | No | Yes |
| Languages | Any language | Python, TypeScript |
| Install | — | pip install rebuff |
| Keys required | LAKERA_GUARD_API_KEY | None |
| Job within the layer | injection | injection |
The deciding difference is operational: Rebuff can run on infrastructure you control, Lakera Guard cannot. If that constraint is real for you, it settles the choice before anything else is considered. Lakera Guard: Prompt-injection and jailbreak detection as a low-latency API call. Rebuff: Open-source prompt-injection detector with canary tokens.
Lakera Guard is a hosted service only. Rebuff can run on your own infrastructure.
Lakera Guard is proprietary (freemium). Rebuff is open source (open source).
Every page here answers to Accept: text/markdown and returns the same content at roughly a tenth the tokens. No separate site, no toggle — same URL.
curl -s -H "Accept: text/markdown" https://newagent.build/compare/lakera-vs-rebuff