In its latest weekly roundup, AWS announced the availability of the GPT‑6.1 Terra model on the Bedrock service, new agent‑capable Step Functions, and added post‑quantum cryptographic key support to its AI infrastructure.

AWS has expanded its AI portfolio, adding the GPT‑6.1 Terra model to Bedrock, introducing agent‑enabled Step Functions, and enabling post‑quantum cryptographic keys for its AI services.

GPT‑6.1 Terra on Bedrock

The new GPT‑6.1 Terra model is now available through Amazon Bedrock, allowing developers to access a higher‑capacity, multimodal language model without managing underlying infrastructure.

Terra builds on the previous GPT‑6 series with improved reasoning, larger context windows, and native support for image and audio inputs, positioning it as a direct competitor to other leading foundation models.

Agent‑enabled Step Functions

AWS Step Functions now include an agent tool integration, enabling state machines to invoke AI agents that can make decisions, call external APIs, and orchestrate complex workflows autonomously.

  • Define agent actions directly in the state definition
  • Leverage Bedrock models for natural‑language reasoning
  • Combine with existing Lambda and ECS tasks for hybrid workflows

Post‑quantum key support

To address emerging security concerns, AWS added post‑quantum cryptographic key options for data at rest and in transit within its AI services, aligning with industry standards for future‑proof encryption.

Customers can now select lattice‑based key algorithms when configuring Bedrock endpoints, ensuring that model outputs remain protected against quantum‑capable adversaries.

For more details on how these features integrate with existing AWS infrastructure, see the Infrastructure category on uxc.news.

Company updates and strategic implications are tracked on Amazon.com, Inc. on Momentum.

Read the full announcement in the AWS blog post on the weekly roundup.