AWS ML ★ 81 3 min

Introducing explicit prompt caching for OpenAI GPT-5.6 models on Amazon Bedrock

Advanced (300)Amazon BedrockTechnical How-to

🔗 https://aws.amazon.com/blogs/machine-learning/introducing-explicit-prompt-caching-for-openai-gpt-5-6-models-on-amazon-bedrock/

📌 【Amazon Bedrock 新功能】GPT-5.6 系列正式登場,顯式 Prompt Caching 讓 Agent 成本直降九成

TL;DR:OpenAI GPT-5.6 系列登陸 Amazon Bedrock,新增「顯式 Prompt Caching」功能,重複使用的快取輸入可享 90% 折扣。

OpenAI 最新的 GPT-5.6 系列模型現在已經在 Amazon Bedrock 上正式可用。這次更新不僅帶來了全新的模型家族,更引入了「顯式 Prompt Caching(Explicit Prompt Caching)」功能,讓開發者能精準控制哪些 Prompt 內容需要被快取,大幅優化 Agentic Workflow(代理型工作流)的成本結構。

🧩 GPT-5.6 家族:針對不同需求提供三種能力層級

這次在 Amazon Bedrock 上推出的 GPT-5.6 模型家族,根據任務複雜度分為三個層級:

  • GPT-5.6 Sol:專為最複雜的推理(Reasoning)與 Agentic Coding(代理編程)工作設計。
  • GPT-5.6 Terra:適合日常生產環境中,兼顧效能與成本的平衡型工作負載。
  • GPT-5.6 Luna:針對高吞吐量、快速任務(如分類與摘要)進行最佳化。

💡 顯式 Prompt Caching:精準控制快取內容並節省 90% 成本

以往的快取機制通常是隱式的,而這次 GPT-5.6 引入的「顯式 Prompt Caching」讓開發者擁有更精確的控制權。

  • 節省成本:使用快取的輸入內容(Cached input),其費用僅為原來的 10%(即 90% 折扣)。
  • 快取時效:快取內容會保留 30 分鐘,供後續請求重複使用。
  • 最佳適用場景:在 Agentic Workflow 中,系統指令(System instructions)、工具定義(Tool definitions)以及參考文件(Reference documents)通常會在多次呼叫中重複出現,使用此功能能獲得最大價值。

🛠️ 開發者整合指南:使用 OpenAI 相容的 Responses API

開發者可以透過 Amazon Bedrock 的 bedrock-mantle 端點,使用與 OpenAI 相容的 Responses API 來呼叫這些模型。

  • 身份驗證:建議使用由 AWS 憑證產生的短期 Bearer Token。
  • 實作方式:開發者可以在安裝 OpenAI SDK 的同時,安裝 Token 產生器,並利用標準的 AWS 憑證鏈(如 IAM Roles 或環境變數)來建立 Client。

🎯 實務啟示

對於正在開發 Agent 或需要處理大量長文本(Long Context)的工程師來說,這是一個關鍵的成本優化手段。透過手動指定需要快取的 Prompt 區塊,可以大幅降低在重複對話或複雜代理任務中的 Token 消耗支出。

🔗 來源

#OpenAI #AmazonBedrock #GPT56 #PromptCaching #GenerativeAI #LLM #AWS #MachineLearning #AIEngineering #AgenticWorkflow

原始資料 AWS ML · 收集於 2026-07-31
來源原標題
Introducing explicit prompt caching for OpenAI GPT-5.6 models on Amazon Bedrock
作者
Melanie Li
原始標籤
Advanced (300) · Amazon Bedrock · Technical How-to
原始連結
https://aws.amazon.com/blogs/machine-learning/introducing-explicit-prompt-caching-for-openai-gpt-5-6-models-on-amazon-bedrock/

摘要原文

This post is co-written with Chris Dickens from OpenAI. OpenAI GPT-5.6 Sol, Terra, and Luna are now generally available on Amazon Bedrock . With GPT-5.6 on Amazon Bedrock, you get the newest generation of OpenAI frontier models with pay-per-token pricing, AWS security and governance controls, and usage that counts toward your existing AWS commitments. The family covers three capability tiers: GPT-5.6 Sol for the most complex reasoning and agentic coding work, GPT-5.6 Terra for balanced everyday production workloads, and GPT-5.6 Luna for fast, high-volume tasks such as classification and summarization. Alongside the new models, GPT-5.6 introduces explicit prompt caching on Amazon Bedrock, a new capability that gives you precise control over which portions of your prompt are cached and reused across requests. Cached input is billed at a 90 percent discount (see the Amazon Bedrock pricing page ) and stays available for reuse for 30 minutes. You get the most value from this in agentic workflows, where system instructions, tool definitions, and reference documents repeat across many calls. In this post, we introduce explicit prompt caching for GPT-5.6 on Amazon Bedrock, a new capability that gives you control over which parts of your prompt are cached and reused across requests. We show how it differs from implicit caching and when to use each mode, how to set it up and verify that it is working, and how it performs in an agentic workflow. We also cover how to migrate workloads from earlier GPT models onto GPT-5.6, whether they run on Amazon Bedrock today or on another platform. GPT-5.6 models are served through the OpenAI-compatible Responses API on the Amazon Bedrock bedrock-mantle endpoint. The recommended way to authenticate is with short-term bearer tokens generated from your AWS credentials. Install the token generator alongside the OpenAI SDK: Then create a client. The token generator uses the standard AWS credential chain (IAM roles, environment variables, or you

tencent/hy3:free 自動生成