OpenAI's GPT-6 Astra Moves AI From Answering Questions to Doing the Work

OpenAI's GPT-6 Astra is a more consequential business release than a simple model-score upgrade. Introduced on 3 September, Astra combines frontier reasoning with computer use, browsing, coding and document creation, pushing the product proposition from generating an answer toward carrying a multi-step workflow through software to a finished result.

For US companies, that distinction changes the economic question. A model that drafts text competes mainly for knowledge-worker attention. A model that can navigate applications, update records, research across sources, manipulate files and produce usable documents begins to compete for pieces of an operating process.

Computer use is the capability businesses should watch

OpenAI reports a 72.6% score for Astra on OSWorld 2.0, compared with 65.7% for GPT-5.6 Sol, and says the simulated tasks were completed in about 47% less time. On AutomationBench, which tests professional workflows, OpenAI reports 41.4% for Astra versus 18.1% for GPT-5.6 Sol. Benchmark results do not equal production ROI, but the direction is commercially relevant: models are being trained to interact with the same software environments employees already use.

OpenAI also says Astra can create documents, spreadsheets and presentations that follow templates and business instructions. That makes the adoption test more demanding. Enterprises will judge not only whether the model can reason, but whether it can work inside permission boundaries, preserve data controls, recover from errors and produce outputs that require less human cleanup.

Our view: Astra raises the value of workflow ownership

American Commerce Review's view is that the near-term winners will not necessarily be the companies that deploy the most AI seats. They will be the companies that know which workflows can safely be delegated, what a successful result looks like and where a person still has to approve an action.

Astra's cybersecurity capability makes that governance question sharper. OpenAI says the model reaches its Critical cybersecurity threshold and has introduced stronger safeguards and monitoring around deployment. Businesses evaluating Astra should therefore treat capability and control as the same procurement decision. The commercial opportunity grows as the model can do more, but so does the cost of poorly specified permissions.