Artificial intelligence news.
Straight from the original announcements.

GPT-6.1 Sol arrives on Amazon Bedrock, aimed at agentic coding and work

September 29, 2026, 21:34By the AI News Bites Editorial TeamSource: AWS
GPT-6.1 Sol arrives on Amazon Bedrock, aimed at agentic coding and work
Image: AWS

GPT-6.1 Sol is now generally available on Amazon Bedrock, AWS said, bringing stronger reasoning to agentic coding, computer use and professional workloads. According to OpenAI, the model delivers near-Astra intelligence for everyday workflows.

Key points
  • GPT-6.1 Sol is generally available on Amazon Bedrock, per AWS.
  • OpenAI says it matches GPT-6 Astra on DeepSWE v1.1 at roughly one-fifth the cost per task.
  • OpenAI says it beats GPT-6 Sol's best DeepSWE score by 6.4 percentage points at lower reasoning effort.
  • Codex and ChatGPT Work can use the model, and developers can build with supported Bedrock APIs.
  • AWS says inference data is not used for training and runs with zero-operator access.

Reasoning quality and the cost of a task

AWS frames the release around the economics of AI agents. To finish a task, an agent may need to gather information, use tools, test different approaches, recover from errors and verify its result. A wrong turn can add model interactions, tool calls, latency and human intervention.

In that view, token prices set the cost of each interaction, while reasoning quality determines how many interactions a task needs and whether they succeed. The total cost of completing a task depends on both. AWS describes GPT-6.1 Sol as a major upgrade to GPT-6 Sol, running on an inference engine built for performance, security and reliability at scale.

Coding and document work

On software engineering, OpenAI says GPT-6.1 Sol matches GPT-6 Astra on DeepSWE v1.1 at roughly one-fifth the cost per task. According to OpenAI, it also exceeds the best score from GPT-6 Sol by 6.4 percentage points while using a lower reasoning effort than that GPT-6 Sol result.

Codex can be configured to use GPT-6.1 Sol on Amazon Bedrock for investigation, implementation and testing. Codex works with repositories, local files, terminals and development tools to write features, fix bugs and run tests, and is available through the desktop app, CLI and supported IDEs. For AWS development, the Agent Toolkit for AWS connects Codex to AWS documentation, APIs and services through a single terminal command.

For document-heavy work, OpenAI says the model approaches GPT-6 Astra on complex document analysis and improves on GPT-6 Sol in multistep workflows across business tools. In the desktop app, ChatGPT Work can gather information across files and applications and turn it into finished deliverables. Through supported Amazon Bedrock APIs, developers can build internal tools that synthesize documents, agents that coordinate work across systems, and customer-facing applications that evaluate multiple inputs.

Limits, approvals and controls

AWS also highlights how an agent handles failure. It should recognize when a tool has failed, an action is restricted or information is missing, and communicate that so the application or user can step in. According to OpenAI, GPT-6.1 Sol improves on GPT-6 Sol in challenging evaluations of transparency, user intent and explicit restrictions.

When building with supported Amazon Bedrock APIs, developers can define the tools available to the model and decide how the application responds when an action requires approval or cannot be completed. AWS says these application-level controls complement the model improvements by keeping people involved at consequential decisions.

Security, data handling and getting started

Amazon Bedrock offers several controls for production use:

  • Model access can be governed through AWS Identity and Access Management (IAM) policies, and invocations audited through AWS CloudTrail.
  • Virtual private cloud (VPC) endpoints powered by AWS PrivateLink can help keep traffic within network boundaries.
  • Inference runs on hardware-isolated infrastructure with zero-operator access, so AWS operators cannot access prompts and completions during inference.

According to AWS, inference data is not used for model training, and using GPT-6.1 Sol does not require opting into sharing data with OpenAI. For automated abuse detection, classifier-flagged traffic is retained by AWS for up to 30 days and processed programmatically. Customers can request zero data retention through their AWS account team.

Users can get started in the Amazon Bedrock console or programmatically through supported Amazon Bedrock APIs. Supported AWS Regions, endpoints, APIs, features, inference profiles and pricing are listed in the Amazon Bedrock documentation; the announcement itself does not give pricing figures.

What it means for teams building agents

For teams running agents on AWS, the pitch is fewer steps to a correct result rather than a lower token price alone. The performance and cost claims come from OpenAI and relate to specific benchmarks, so results on real workloads may differ. Organizations that need governance and audit features alongside the model can use the existing Bedrock tooling described above.

Related news

OpenAI and Databricks bring governed enterprise data to AI tools
BusinessNovember 13, 2026, 16:30

OpenAI and Databricks bring governed enterprise data to AI tools

OpenAI and Databricks are expanding their partnership to help enterprises use ChatGPT, Codex and GPT models with governed business data. The companies describe three…

Products and tools
Products and toolsNovember 12, 2026, 21:00

Amplitude says OpenAI-powered Wave ships product changes in 1, 2 days

Amplitude says its Wave product agent can help teams move from product insight to shipped improvement in 1, 2 days, down from roughly a month. The agent uses OpenAI…

ChatGPT Desktop App Now Runs on Amazon Bedrock in AWS GovCloud (US)
For developersOctober 30, 2026, 18:27

ChatGPT Desktop App Now Runs on Amazon Bedrock in AWS GovCloud (US)

The ChatGPT desktop app is now available on Amazon Bedrock in AWS GovCloud (US), according to AWS. A new walkthrough shows how organizations can route model requests…