Amazon Bedrock Introduces Rate Limiting for AI Traffic on AgentCore Gateway
Amazon Bedrock has announced the implementation of rate limiting on its AgentCore gateway, a serverless AI gateway that manages AI traffic. This new feature allows for fine-grained control over the amount of traffic individual users can consume, ensuring that downstream services remain available during heavy traffic spikes. The gateway supports three target types: MCP targets, inference targets, and HTTP passthrough targets, with rate limiting metrics applicable to all. The system uses OAuth or IAM-based rules to manage requests per minute, concurrent connections, and token throughput, providing a centralized mechanism for managing AI traffic. 
