GPT-6 Astra Now Generally Available on Amazon Bedrock

What changed: GPT-6 Astra became generally available on Amazon Bedrock, enabling enterprise users to integrate it into existing AWS workflows.

GPT-6 Astra, OpenAI's latest and most capable model, is now generally available on Amazon Bedrock. The model is also accessible via ChatGPT Work, Codex, and the API. With a context window of up to one million input tokens, it can handle large-scale document processing scenarios. GPT-6 Astra supports both implicit and explicit prompt caching; with explicit caching, users can set custom breakpoints to control which context is cached, reducing repeated processing cost and latency.

On the security front, GPT-6 Astra is the first OpenAI model to reach the Critical classification for cybersecurity capability under OpenAI's Preparedness Framework. At this level, automated safeguards monitor misuse in real time and can pause or stop activity that exceeds defined boundaries. On Amazon Bedrock, zero-operator access is enforced at the chip level, meaning even AWS operators cannot access prompts and completions during inference. Every invocation is logged in AWS CloudTrail, and classifier-flagged traffic is retained for up to thirty days for automated abuse detection. Inference data is not used for model training, and using GPT-6 Astra does not require opting into data sharing with OpenAI.

In benchmark evaluations, GPT-6 Astra set a new record on DeepSWE v1.1, achieving a result no previous frontier model had reached—with fewer steps and greater token efficiency. OpenAI's own engineering team reported that switching allocators yielded significantly lower turn latency at the cost of roughly thirty percent higher peak memory use. The model is positioned as state-of-the-art across computer use, browsing, professional work, software engineering, cybersecurity, and science.

For enterprise integration, new plugins for Workday, Navan, and Avalara were introduced alongside this release, extending browser-use capabilities across data analytics, operations, and finance. ChatGPT Work is available through the ChatGPT desktop app for Mac and Windows. Pricing starts at ten dollars per million input tokens and fifty dollars per million output tokens.

Key facts

• GPT-6 Astra is generally available on Amazon Bedrock with a context window extending to one million input tokens.
• It is the first OpenAI model to reach the Critical cybersecurity classification under the Preparedness Framework.
• Pricing is set at ten dollars per million input tokens and fifty dollars per million output tokens.
• GPT-6 Astra set a new record on DeepSWE v1.1 with fewer steps and greater token efficiency than any previous frontier model.
• Zero-operator access enforced at the chip level prevents even AWS operators from accessing prompts and completions during inference.

Watch next: Broad adoption of the Workday, Navan, and Avalara enterprise plugins, and the implications of the Preparedness Framework's Critical classification for other frontier models.

Source date: 2026-09-09T23:21:05.228Z · Verification confidence: 85/100