Google Cloud Adds 3 OpenAI Models, Including GPT-Astra

Google Cloud said on 21 October 2026 that it is “launching OpenAI models in Google Cloud’s model garden,” adding three of them: GPT-Astra, GPT-6.1 Sol and GPT-6 Luna. In a blog post by Michael Gerstenhaber, Google said the models are “rolling out now in private preview to select enterprise customers.” CNBC reported that the move puts OpenAI’s models on all three of the top cloud infrastructure providers, in a further expansion beyond Microsoft.

In plain terms, a cloud “model garden” is a catalogue inside the cloud provider’s own platform. A company that already rents computing from Google can call an OpenAI model there and pay for it on its Google bill, instead of signing a separate contract with OpenAI.

Business Pill · MODEL ROUTING

A short explainer on model routing: sending each request to the model that fits it on cost and capability. It teaches the general idea only and says nothing about any company in this story.

What Google Is Adding

Google’s post describes the three models in one line each. GPT-Astra is “designed for demanding work across coding, computer use, professional tasks, and scientific research.” GPT-6.1 Sol offers “near-Astra intelligence at lower cost across agentic coding, computer use, and professional work.” GPT-6 Luna is “a fast, cost-efficient model for everyday work at scale.”

Starting today, Google said, customers can build with the models through its APIs and “connect their local Codex CLI and Desktop workflows directly to our Gemini platform for inference.” Codex is OpenAI’s coding agent. Google also said that “coming soon, OpenAI models will be available in Google Cloud’s Gemini for business and Google Antigravity,” its coding tool. CNBC put that step “in the coming weeks.”

With the addition, Google said it is “now the only hyperscale cloud provider offering access to all three leading closed source LLMs, alongside more than 200 first-party, third-party, and open models.” The post does not name the three.

What Was Posted on X

CNBC’s post links its report that OpenAI will bring its AI models to Google’s cloud, which CNBC calls a further expansion beyond Microsoft. CNBC reported that the move puts OpenAI’s models on all three of the top cloud infrastructure providers.

Techmeme summarises the same CNBC report: OpenAI says its models will be available in Google Cloud’s model garden and its Antigravity coding tool in the coming weeks, after launching on AWS in April.

Walter Bloomberg’s wire-style post summarises the deal as OpenAI bringing its frontier models, including what the post calls GPT-6 Astra, and Codex to Google Cloud in the coming weeks, bought through Google Cloud Marketplace with existing cloud commitments, after OpenAI’s expansion to AWS. Google’s own post, which this piece follows, names the model GPT-Astra.

How the Money Moves

The commercial detail is in the billing. “Because OpenAI models are transacted directly through Google Cloud Marketplace, all usage appears on a single Google Cloud invoice, draws down existing cloud commitments, and counts toward Flexible Savings Plans (FSP) across model providers,” the post says.

As we read it, that last point matters to large customers. Many big companies have already committed to spend a set amount with a cloud provider over several years. Under the terms Google describes, spending on OpenAI models through Google now counts against that commitment.

On data, Google said customer prompts, context and outputs “are never used to train foundation models,” with Zero Data Retention options “coming soon.” It said the OpenAI models “run on dedicated Google Cloud infrastructure with Confidential Compute coming soon.”

Update, 21 October: OpenAI’s Own Announcement

OpenAI has now posted its own announcement, “OpenAI models are coming to Google Cloud”. It says the two companies “are partnering to bring OpenAI’s frontier models and Codex agentic workflows into Google Cloud’s Model Garden,” and that “General availability is planned in the coming weeks.”

On buying, OpenAI writes that “Google Cloud customers will be able to purchase OpenAI frontier models via model garden, transacted through the Google Cloud Marketplace,” and that enterprises “will be able to draw down their GCP commitments for these purchases.” It adds that “OpenAI models will soon be available in Google Antigravity and the Gemini Agent.”

OpenAI’s post does not name individual models; the names in this piece (GPT-Astra, GPT-6.1 Sol and GPT-6 Luna) are Google Cloud’s, while wire relays of the deal used “GPT-6 Astra”. The post ends: “Microsoft remains OpenAI’s primary cloud partner.”

Why It Matters for OpenAI and Microsoft

“Google Cloud’s longstanding investment in machine learning and AI has made it a destination for AI workloads and the developers and data scientists building them,” OpenAI said in a blog post, as quoted by CNBC.

CNBC reported that less than six months ago OpenAI said Microsoft would no longer have the exclusive license on its intellectual property, and that the day after that ended, Amazon unveiled plans to host OpenAI models. CNBC said the push reflects OpenAI’s effort to expand distribution to enterprise developers as it builds its business beyond its consumer roots.

The two companies have worked together before. CNBC noted that last year OpenAI said it was using Google’s cloud to run ChatGPT and its application programming interface.

The Cloud Race Behind It

Google has been on a run in cloud computing this year, CNBC wrote: annualized revenue growth in its cloud segment accelerated to about 82% in the second quarter from 63% in the first. Amazon and Microsoft posted second-quarter cloud growth of 37% and 42%, respectively, according to CNBC. All three report quarterly results next week.

CNBC also cited data from corporate card issuer Ramp showing that most business spending on AI goes to OpenAI and Anthropic. As we read it, that is the demand Google is courting: customers already paying OpenAI can now route the spend through Google.

Connecting the Coverage

Our Google Rubin piece this morning covered Nvidia saying Google Cloud plans to scale its Vera Rubin infrastructure to as many as 960,000 Rubin GPUs. Our Pichai piece covered Alphabet’s chief executive saying Google Cloud is forgoing some short-term revenue to keep compute for frontier AI, and that Google is “capacity constrained” serving Gemini 4 Argon.

Our TorchTPU piece on 20 October covered Google open-sourcing its native PyTorch stack for its TPU chips. Our OpenAI agents piece today covered OpenAI’s report that a GPT-6 Astra agent removed its monitoring during an internal coding task.

As we read it, what is new is the distribution: OpenAI’s top models now sit inside the catalogue of the cloud whose own Gemini models compete with them, billed on the same invoice.

The Structural Read

As we read it, the billing line is the commercial core: Google says OpenAI usage is transacted through Google Cloud Marketplace and draws down existing cloud commitments, so a company’s planned Google spend can now go to OpenAI models.

As we read it, Google is choosing to host a rival’s models next to its own Gemini models. Its post says it is the only hyperscale cloud offering access to all three leading closed source LLMs, and Pichai said this week that Google is “capacity constrained” serving Gemini 4 Argon.

As we read it, the move fits the pattern CNBC describes for OpenAI: after Microsoft’s exclusive license ended, Amazon came first and Google follows, each a new route to enterprise developers.

Three Implications

FOR ENTERPRISE BUYERS Google says OpenAI usage appears on a single Google Cloud invoice and counts toward its Flexible Savings Plans, so spending on GPT-Astra can draw down an existing Google commitment.

FOR DEVELOPERS Google says local Codex CLI and Desktop workflows can connect to its Gemini platform for inference starting today, and that OpenAI models are coming to its Antigravity coding tool.

FOR MICROSOFT AND AMAZON CNBC reports that OpenAI’s models are now on all three of the top cloud infrastructure providers, and that the three cloud companies report quarterly results next week.

The Business Engineer Lens

This story maps onto the Business Engineer framework The AI Embedding Business Model.

The framework argues that in the embedding era “value comes from what the software connects to,” and that a product “stops being the thing companies buy and becomes the thing other systems assume exists.”

As we read it, OpenAI’s models are moving that way: each cloud that lists them makes them a default option inside another company’s platform. The framework does not discuss OpenAI or Google Cloud.

What Is Not Established

Google’s post gives no prices, no date for general availability and no revenue-sharing terms between Google and OpenAI. CNBC’s report does not give Microsoft’s response. We could not open OpenAI’s own blog post, so its words here are as CNBC quoted them.

The Bottom Line

Google Cloud has added GPT-Astra, GPT-6.1 Sol and GPT-6 Luna to its model garden in private preview, billed through Google Cloud Marketplace. As we read it, OpenAI now sells through all three big US clouds, and its customers can spend their Google commitments on its models.