AITOPAI热报
ProductsOfficialSelected

Amazon Bedrock adds prompt caching

Optimizing cost and latency with Amazon Bedrock prompt caching

官方一手AWS Machine Learning Blog@Daniel Abib
Blog/Media
Read the original
Why we picked it

For developers using Amazon Bedrock, this feature can significantly reduce costs for repeated API calls.

Amazon Bedrock now supports prompt caching, which can reduce input token costs by up to 90% for repeated requests. The feature is demonstrated through six practical scenarios using the Converse API, including message content, system prompts, and tenant isolation.

Image · AWS Machine Learning Blog
Original · AWS Machine Learning Blog

Optimizing cost and latency with Amazon Bedrock prompt caching

Prompt caching in Amazon Bedrock can cut input token costs by up to 90% when you repeatedly send the same context to foundation models. This post walks through six practical prompt caching scenarios using the Converse API: message content, system prompt, tool definition, mixed TTL, tenant isolation, and LangChain integration.

#Amazon Bedrock#prompt engineering#cost optimization#AI products

Keep reading

Recent stories with the same tag or section
ProductsBuildersOfficialSelected05:58
Amazon Bedrock AgentCore adds live web preview for coding agents

Live preview closes the feedback loop that normally breaks when a coding agent works inside a sandbox: developers can watch rendered output change on each edit rather than diffing code. Teams building agent workflows on AWS can use session endpoints to expose any port-based tooling, not just websites. Against alternatives like cloud IDE previews, the advantage is per-session authenticated URLs tied directly to the agent runtime.

#Amazon Bedrock#AgentCore#coding assistants#agents
官方一手AWS Machine Learning Blog@Sundar RaghavanSource
ProductsBuildersOfficialSelected01:13
MoEngage turned 400 APIs into agent tools with AgentCore Gateway

The key insight is that a target-resolution layer plus Cedar policies can make a large API fleet usable by agents without flooding model context with 400 tool definitions. Engineering teams exposing internal APIs to LLM agents can replicate this pattern with AgentCore Gateway instead of hand-rolling per-service MCP servers. It is a concrete reference for enterprises on AWS managing multi-region private endpoints.

#MoEngage#AgentCore Gateway#Amazon Bedrock#MCP
官方一手AWS Machine Learning Blog@Nikhil KumarSource
How-toBuildersOfficialSelected01:18
Fix evaluation bias to unlock better models on Amazon Bedrock

Labels inherited from the model you are replacing quietly lock in its errors, so benchmark gains can be illusory. Engineering and product teams running model evaluations can reuse this method to validate switching decisions with independent ground truth. The 86% cost reduction also shows the payoff when a fairer comparison reveals cheaper models that were previously scored unfairly.

#Amazon Bedrock#Property Finder#prompt engineering#evaluation
官方一手AWS Machine Learning Blog@Khaled ThabitSource
How-toBuildersOfficialSelected01:32
AWS CDK template builds agentic retrieval on Amazon Bedrock Knowledge Bases

Routing across multiple knowledge bases through one gateway reduces the glue code teams usually write by hand for enterprise RAG. Builders evaluating agent frameworks can reuse the CDK stack to get observability and evaluation pipelines from day one instead of assembling them separately. It sits alongside alternatives like LangChain-based retrievers, with the advantage of managed Bedrock infrastructure.

#Amazon Bedrock#AWS CDK#RAG#agents
官方一手AWS Machine Learning Blog@Prabhakar ChandrasekaranSource