OpenAI GPT-6 models come to Google Cloud in preview

OpenAI's GPT-6 models are coming to Google Cloud, the two companies announced on October 21, 2026. GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna are rolling out in private preview to select enterprise customers. For teams already on Google's cloud, that means OpenAI models on the same bill, under the same contracts.

What Google Cloud is offering

The models live in Model Garden, Google's catalog of AI models that customers can call through its APIs. The Google Cloud announcement describes each one in a line.

Model	Google's description
GPT-6 Astra	"Designed for demanding work across coding, computer use, professional tasks, and scientific research"
GPT-6.1 Sol	"Near-Astra intelligence at lower cost across agentic coding, computer use, and professional work"
GPT-6 Luna	"A fast, cost-efficient model for everyday work at scale"

Customers buy the models through Google Cloud Marketplace. Google says use appears on the one Google Cloud invoice and can draw down existing spending commitments. It also counts toward Flexible Savings Plans, Google's discounts for committed spend across model providers.

Google publishes no prices and no date for general availability. OpenAI plans general availability "in the coming weeks," The Next Web reports. Interested customers can apply through an interest form on Google's site.

Codex and the data rules

The deal also covers Codex, OpenAI's coding agent. Starting now, Google says, developers can connect local Codex CLI and Desktop workflows to Google's Gemini platform for inference. Inference is the work of running a model to produce an answer.

Google says prompts, context and outputs stay inside each customer's chosen data boundary. They are not used to train foundation models. Two stricter options, Zero Data Retention and Confidential Compute, are listed as "coming soon."

OpenAI models will also come to Gemini for business and Google Antigravity, Google's coding tool, according to Google. The Next Web names Antigravity and the Gemini Agent instead.

Why OpenAI is adding a third cloud

Microsoft remains OpenAI's primary cloud partner, OpenAI says, according to The Next Web. Amazon Web Services has hosted OpenAI's models since April, after Microsoft's exclusive license ended. Tech AI Wire covered that in GPT-6 Astra is generally available on Bedrock and Copilot.

Google, in turn, now claims to be the only large cloud provider offering Gemini, Claude and GPT models side by side. Its Model Garden lists more than 200 first-party, partner and open models.

OpenAI president Greg Brockman called the move "an important step in making our most capable models and Codex available where customers are already building." Google Cloud chief executive Thomas Kurian said the company is "focused on giving customers a broad selection of models." Analyst Mike Leone of Moor Insights & Strategy told Axios: "Gemini now has to win those customers with OpenAI sitting right next to it."

What this means for developers

For most teams, this is a procurement change before it is a technical one.

• Check your committed spend. If your company has a Google Cloud commitment, GPT-6 calls can draw it down. That lets you spend money already committed to Google instead of opening a new OpenAI contract.
• Do not plan production on the preview. Access is limited to select enterprise customers, and there is no published price yet.
• Wait if you need zero retention. Zero Data Retention is not available yet on Google Cloud. Regulated workloads that require it should wait.
• Compare prices once they appear. GPT-6 Astra already runs on Amazon Bedrock, which also hosts OpenAI agents, as covered in Amazon Bedrock Managed Agents for OpenAI enters preview. The same model on three clouds gives you room to negotiate.
• Try the Codex link if you live in Google Cloud. Routing local Codex sessions through Google keeps the usage on one bill and inside one set of access controls.