Computer use
Filling forms, gathering research, running frontend QA — browser and desktop work done faster and more reliably.
A new generation of intelligence · Computer use / Coding / Professional work
OpenAI's most intelligent and aligned model yet, setting a new frontier on computer use, coding and professional work.

Model overview
GPT-6 Astra is OpenAI's new flagship, which they call their most intelligent and aligned model: state of the art on computer use, browsing, software engineering, cybersecurity and science, and faster and more accurate on demanding professional work.
It is better at seeing long tasks through. When instructions leave room for interpretation it fills routine gaps itself and asks only where the answer would change the outcome, and a mid-course correction no longer drops the original constraints. In Codex it can keep notes across context windows, so long sessions stop losing detail to repeated compaction — and it still retrieves earlier requirements reliably at million-token context lengths.
Through XylemNode, a single API key lets you call it directly from editors, the command line and any client — with no separate OpenAI integration.
Key capabilities
Filling forms, gathering research, running frontend QA — browser and desktop work done faster and more reliably.
Documents, spreadsheets and decks that follow your templates and carry only the context that matters.
Still finds earlier requirements and conclusions at million-token lengths, so long tasks stay on track.
Specs
Bring this model into your workflow