Amazon Bedrock Guardrails announces an increase in service quota limits

Amazon Bedrock Guardrails announces an increase in default service quota limits enabling you to scale your generative AI applications for higher traffic. Bedrock Guardrails provides configurable safeguards to filter undesirable and harmful content across different categories and prompt attacks, topic filters to define and disallow specific topics, sensitive information filters to redact personally identifiable information (PII), word filters to block specific words, and detect model hallucinations by detecting grounding and relevance of model responses as well as identify, correct, and explain factual claims in model responses using Automated Reasoning . These policies can be tailored to your specific use cases and responsible AI policies. Guardrails can be applied across any foundation model including those hosted with Amazon Bedrock, self-hosted models, and third-party models using the ApplyGuardrail API, providing a consistent user experience and standardizing safety and privacy controls.

Starting today, Bedrock Guardrails enables you to scale your generative AI applications for higher traffic loads with increased service quota limits that help process higher transactions per second (TPS) and higher text units per second (TUPS). With this increase, you can now process up to 50 calls per second using the ApplyGuardrail API, a 2x increase from the previous limit of 25 calls per second. Content filters, sensitive information filters, and word filters can now process up to 200 TUPS, a 8x increase from the previous limits of 25 TUPS.

These limits are available in US East (N. Virginia) and US West (Oregon) AWS regions.

To learn more, see the technical documentation and the Bedrock Guardrails product page.