Industry story
Amazon Bedrock Launches Built-in Web Search for OpenAI GPT Models
agent-framework cloud-costs orchestration tool-use
Amazon just made it easier to build agents on Bedrock by folding web search directly into the platform. GPT-4.5, GPT-4.6, and GPT-4.7 models now get real-time grounding through a single API parameter, no third-party search keys, no separate vendor contracts, no custom orchestration. For agent builders, that removes a meaningful chunk of the plumbing. The stickier question is whether consolidating retrieval inside AWS is a simplification or just a different kind of lock-in.
Full analysis
AWS has made Web Search on Amazon Bedrock generally available, offering a server-side web grounding tool ("grounding" means anchoring AI responses to real-world, up-to-date information) that works natively with OpenAI's GPT-5.4, GPT-5.5, and GPT-5.6 models without requiring third-party search providers. The feature eliminates the previous need for external API keys, separate vendor onboarding, custom orchestration, and compliance reviews by handling the entire search lifecycle within AWS via a single API parameter. Amazon's index spans tens of billions of documents refreshed continually, includes a built-in knowledge graph, and performs semantic snippet extraction optimized for model context windows. It integrates through a standardized tool-use interface compatible with the OpenAI Responses API. The launch is significant for agent builders because web search (real-time retrieval) is a core tool-use capability, and having it natively in-platform reduces infrastructure complexity, data-egress risk, and vendor sprawl. It is available now in three US AWS regions.
Comments