F5 AI Guardrails
#4 in Observability & EvalsF5 · 2× · tolest 07. Aug. 2026
88
Momentum
F5 AI Guardrails is a Runtime Security solution from F5 that protects AI models and agents from Prompt Injection, Jailbreaks, data leaks, and malicious outputs. The solution is model-agnostic, operates at the Inference layer between applications and AI models, and offers compliance controls (such as GDPR, HIPAA, EU AI Act, PCI, PHI) as well as Audit-Logging and an Observability Dashboard. It is part of the F5 Application Delivery and Security Platform (ADSP), was created following the acquisition of CalypsoAI, and has been generally available since January 2026; in July 2026, an integration with NVIDIA NeMo Guardrails was added.
Momentum-Verloop
20.05.18.08.
Features
| Deployment (Self-Hosted/Cloud) | SaaS or self-hosted on AWS EKS, Azure AKS, Google GKE, or on-prem via Red Hat OpenShift |
| Platform | Part of the F5 Application Delivery and Security Platform (ADSP) |
| Price | No public pricing; custom quote via F5 sales ('Private Offer'), subscription-based on usage/apps/users |
| Protocol Compatibility | Inline inspection of prompts/responses via API endpoint integration, no application code changes required |
| Release Date | General availability (GA) in January 2026 |
| Supported Models/Providers | Model-agnostic, supports millions of public and proprietary models, independent of cloud/model provider |