Top News

Microsoft CEO Satya Nadella has confirmed that early customers are already using GPT-6 Astra on Azure, offering an early indication of how quickly OpenAI’s newest frontier model is moving from launch into enterprise environments. Nadella shared the update on X on September 4, writing that he was “excited to see early customers already using Astra on Azure.” His post linked to Microsoft’s announcement detailing the model’s availability through Microsoft Foundry.

The update is significant because Astra is not being positioned simply as another chatbot model. Microsoft describes it as a system designed to help organizations reason through complex work and execute tasks across applications and systems.

GPT-6 Astra Begins Its Azure Rollout

Microsoft announced on September 3 that GPT-6 Astra had begun rolling out through the Microsoft Foundry Limited Access Program, with availability expanding to participating customers. The initial rollout means access is still limited rather than being a general release for every Azure customer.

Microsoft is using Foundry as the enterprise platform through which organizations can evaluate and deploy the model while managing identity, networking, governance, data handling, and compliance requirements. Nadella’s latest update now provides confirmation that the rollout has already reached customers.

GPT-6 Astra: Replit Among the Early Customers

Microsoft has highlighted Replit as one of the organizations using GPT-6 Astra through Microsoft Foundry. Luis Hector Chavez, CTO of Replit, said the model opens up a new level of agentic capability beyond traditional code generation and can support active software creation. The example shows how Astra is being positioned for workflows in which AI can participate more directly in building and completing tasks.

Microsoft also cited Albertsons Companies in its announcement, highlighting its use of Azure OpenAI on Microsoft Foundry to balance access to advanced AI capabilities with enterprise security, governance, and operational controls.

Astra Targets Complex Enterprise Work

Astra is designed to take an open-ended challenge, reason through multiple steps, create a plan, and produce a finished result. It can also incorporate new instructions during a task and use tools across applications and systems. This gives the model a broader role than conventional conversational AI. Instead of simply responding to a question, Astra can help carry out larger units of work, including preparing documents, spreadsheets, presentations and analyses for human review.

Microsoft is particularly highlighting use cases in software engineering, business intelligence and professional workflows. For software teams, Astra can help reproduce complex bugs, investigate possible causes, propose fixes and prepare changes for developer testing. In business intelligence, it can help build and refine Power BI dashboards and prepare insights for review.

Computer Use Brings Another Layer of Automation

One of Astra’s notable capabilities is computer use. Microsoft says the model can interpret information displayed on a screen and interact with approved software interfaces, including workflows where dedicated APIs are unavailable.

That can allow Astra to support tasks such as updating records, navigating development tools, testing websites, and assembling information into reports. Microsoft stresses that these workflows require appropriate permissions, approvals, monitoring, and human oversight, particularly when an AI agent is performing consequential actions.

How Enterprise Controls Remain Important

As Astra moves into enterprise environments, Microsoft is also emphasizing the controls surrounding its deployment. Microsoft Foundry provides capabilities including identity and access management through Microsoft Entra, encryption, private networking options, role-based access controls, content filtering, safety evaluations, monitoring, and governance tools.

Microsoft also states that prompts and outputs are not used to train the models. The company notes, however, that these controls do not remove the risks of agentic AI and that organizations remain responsible for configuring safeguards appropriate to their own use cases.

A Step Toward Agentic Enterprise AI

The Azure rollout also reflects a broader shift in how Microsoft and OpenAI are approaching enterprise AI. Instead of treating advanced models purely as assistants that generate text or answer questions, the focus is increasingly moving toward systems that can participate in complete workflows. For businesses, that distinction could influence how AI is integrated into everyday operations.

Teams may increasingly use AI alongside existing software rather than relying on separate chat interfaces for individual tasks. The early Azure deployment therefore serves as an important testing ground for this model of enterprise computing, where AI can become part of a workflow while employees continue to provide oversight and make final decisions.

Availability of GPT-6 Astra

GPT-6 Astra is currently rolling out through the Microsoft Foundry Limited Access Program. Microsoft says it will be available through Standard deployments, including Global and U.S. Data Zone options. The service uses consumption-based pricing, with costs varying according to deployment type, context length, and usage. Microsoft has published separate pricing for Global and U.S. Data Zone deployments.

Conclusion

Satya Nadella’s update confirms that GPT-6 Astra has already begun reaching customers through Azure, giving the model an early foothold in enterprise AI workflows. With Microsoft Foundry providing the deployment and governance layer, the initial customer rollout could offer an early indication of how organizations will use increasingly capable AI systems for complex tasks that extend beyond conventional chat.