GPT-6 Astra in Microsoft Foundry: Availability, Specs and Enterprise Access

Microsoft’s current position is clearer than the early limited-access framing: GPT-6 Astra is generally available to all customers in Microsoft Foundry. Organizations can deploy the model through either Standard or Provisioned Throughput options in Global and US Data Zone geographies. 6

That makes Azure an enterprise distribution channel for Astra, not the model’s exclusive home. OpenAI has also said the rollout includes its API and AWS Bedrock, alongside access for ChatGPT Plus, Pro, Business, and Enterprise users. 16

What enterprises can deploy through Foundry

Microsoft Foundry offers two deployment choices for GPT-6 Astra:

• Standard: pay-as-you-go capacity intended for variable demand.
• Provisioned Throughput: dedicated model-processing capacity for workloads that need more predictable latency and guaranteed throughput.

Both are available in Global and US Data Zone geographies, according to Microsoft’s announcement. 6

For an organization, the practical decision is less about gaining exclusive access to Astra and more about choosing the operational model that fits its workload: flexible consumption for uneven usage, or reserved capacity for sustained production demand.

GPT-6 Astra specifications

OpenAI’s published model comparison lists a 1,050,000-token context window and a 128,000-token maximum output for GPT-6 Astra. 1

Those limits matter for applications that need to work over unusually large bodies of material or produce long structured outputs. They are capacity limits, however—not a guarantee that every long-context request will be interpreted perfectly or that a maximum-length output is appropriate for every task.

OpenAI positions Astra for complex reasoning, coding, research, and computer-use work—tasks that can involve multiple steps rather than a single prompt-and-response exchange. 16 The company’s description should be read as product positioning; the provided primary material does not independently establish a precise definition of autonomous computer use or superiority on a specific benchmark.

The safety claims that remain unverified

Some accounts have attached consequential claims to Astra, including a formal “Critical” cybersecurity designation, independent zero-day-exploit capability, and a launch delay caused by extensive safety testing.

The provided primary OpenAI and Microsoft sources do not substantiate those claims. Without a model card or direct safety report that documents them, they should be treated as unverified rather than as established facts. The same caution applies to broad descriptions of Astra as the industry’s “most capable” or “most aligned” model: those are vendor claims, not conclusions demonstrated by the source material available here.

Azure’s role in the distribution strategy

Astra’s Foundry availability places a frontier model inside Microsoft’s cloud platform and commercial delivery model. Microsoft reported that Azure revenue surpassed $100 billion in fiscal 2026, growing 41% for the full year; Azure and other cloud services grew 43% year over year in the fiscal fourth quarter. 31

That scale helps explain why Azure is an important route for enterprise AI adoption. Foundry gives customers a Microsoft-managed deployment path and a choice between on-demand and provisioned throughput. At the same time, OpenAI’s direct API and AWS Bedrock availability mean enterprises are not limited to a single cloud route. 616

Azure vs. OpenAI API vs. AWS Bedrock

The available channels serve overlapping but distinct procurement and deployment preferences:

Access route	What the announcements establish
Microsoft Foundry	GPT-6 Astra is generally available with Standard and Provisioned Throughput deployments in Global and US Data Zone geographies. 6
OpenAI API	OpenAI identifies its API as a rollout channel for Astra. 16
AWS Bedrock	OpenAI also identifies AWS Bedrock as an availability route. 16
ChatGPT plans	OpenAI says Astra is becoming available to ChatGPT Plus, Pro, Business, and Enterprise users as the rollout expands. 16

The key takeaway is that Microsoft is using Foundry to make Astra consumable through Azure’s enterprise platform, while OpenAI and AWS extend access through their own products. For buyers, the relevant comparison is therefore deployment model, region availability, capacity needs, and existing cloud commitments—not whether the underlying model is available from only one provider.

Bottom line

GPT-6 Astra has moved beyond the premise of a Foundry-only limited-access rollout: Microsoft says it is now broadly available in Foundry, with on-demand and provisioned-capacity options. 6 Its documented headline specifications are a 1.05 million-token context window and a 128,000-token output ceiling. 1

Astra is part of a multi-channel rollout across Microsoft Azure, OpenAI’s API, AWS Bedrock, and eligible ChatGPT plans. 16 Enterprises evaluating it should separate those documented availability and specification facts from unverified safety and exploit-capability claims.