OpenAI's GPT-5.6 family on Bedrock
The smartest family of OpenAI models is now available on Amazon Bedrock, on Bedrock's next-generation inference engine built for high performance, security, and reliability. Three model tiers cover the use case spread. Sol is the flagship reasoning model. Terra is the balanced-performance tier. Luna is the fast, cost-efficient inference tier. All three are accessible through the Responses API on Amazon Bedrock.
This is the first time the GPT-5.6 generation has been available through a managed service at AWS. The arrival matters for teams that have standardized on Bedrock for model access, since the previous OpenAI generations were either direct-API only or available through limited partner programs.
S3 Standard-IA and One Zone-IA: same-day transition
You can now transition objects to S3 Standard-Infrequent Access and S3 One Zone-Infrequent Access as soon as the day they are created, without the previous 30-day minimum retention period in S3 Standard. The two IA storage classes offer up to 40% lower storage costs than S3 Standard while still providing millisecond access when needed.
The change matters most for backup, log analytics, and compliance workloads where data becomes cold within hours or days. The previous 30-day minimum was a structural penalty for these patterns, and the same-day transition option removes it.
One-click Lambda agent setup
The headline of the week is a one-click agent setup prompt for Lambda. The prompt configures your AI agent with AWS Serverless skills and the Serverless Model Context Protocol server, embedding serverless best practices from the start. The prompt references the Lambda agent setup guide, which includes installation commands for Claude Code, Kiro, Cursor, GitHub Copilot, Codex, Devin Desktop, and OpenCode.
To get started, choose the button on the Lambda console screen or copy directly and paste the URL into your preferred AI agent. The MCP angle is worth noting. By embedding the MCP server into the agent's setup, AWS is treating MCP as the standard way for AI agents to interact with cloud infrastructure, and the Lambda prompt is the first major agent-ready onboarding flow from a hyperscaler.
S3 source buckets for Lambda
With self-managed Amazon S3 buckets for code storage, you can reference source code directly from your own S3 buckets without Lambda creating intermediate copies. The change eliminates code storage limits and reduces function activation time after a function is created or updated, because the copy step is gone.
For teams running large monorepos or organizations with strict data residency requirements, the self-managed bucket approach is a meaningful improvement. The intermediate copy was a real cost, both in storage and in the cold-start latency of the first invocation after a deploy.
Other launches worth noting
Amazon Cognito now supports password hashes in CSV user imports. Previously, imported users had to reset their passwords on first sign-in. Now, you can include password hashes in the CSV import, so users sign in immediately with their existing credentials. The change removes a long-standing friction point for migrations from on-prem identity systems.
The full list of AWS announcements is on the AWS What's New page. Roundup posts catch the highlights, but the announcements page is the authoritative list for the week.