AI Security Guardrails can be deployed in two configurations based on your performance requirements:
- CPU-Based (Standard): Installed locally with the AI Gateway. This is the default configuration.
- GPU-Based (High Performance): Installed the LLM on external Graphics Processing Units (GPUs) for optimized throughput and lower latency.
Upon successful configuration of GPU-based guardrails, the system automatically transitions from CPU-based to GPU-based guardrails. No manual switch is required.