OpenAI will sell its GPT-6 family through Google Cloud's Model Garden, putting its flagship models on all three major US clouds and closing an awkward gap in its distribution strategy.[1][2][3]

The deal, announced Oct. 21, covers GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna. Google Cloud customers can buy them through the Google Cloud Marketplace, with usage drawn from existing cloud commitments instead of a separate OpenAI contract.[1][2]

Three models, one invoice

Astra targets demanding work in coding, computer use, professional tasks and scientific research. Sol sits below it in capability and price, built for agentic coding and office work. Luna is the cheap, fast option for high-volume jobs.[1]

Google says its cloud is now the only hyperscaler carrying all three major closed model families: Gemini, Claude and GPT. The Model Garden catalog lists more than 200 first-party, third-party and open models.[1] That breadth is the pitch. Enterprises want to route each workload to whichever model is cheapest or best that week, without signing new vendor paperwork.

Billing follows the same logic. Purchases run through Google Cloud Marketplace, appear on one invoice and count toward Flexible Savings Plans. For a company that has already committed to spend a certain amount with Google, OpenAI tokens become another line item rather than a new procurement project.[1]

Why OpenAI keeps opening new doors

The Google deal is the third stop in a year of widening distribution. Microsoft lost its exclusive license on OpenAI's intellectual property less than six months before this announcement. Amazon unveiled plans to host OpenAI models the day after that exclusivity ended, in April.[3]

Oracle came next. In June, OpenAI said OCI customers could apply Universal Credits toward its models and Codex. By August, purchases were live through Oracle Marketplace in the United States.[4] Each agreement follows the same shape: meet enterprises inside the cloud account they already pay for.

The motive is partly financial. CNBC reports that OpenAI is courting investors at a $1.4 trillion valuation while targeting an IPO.[3] A company preparing to go public needs revenue that does not depend on one cloud partner or on ChatGPT subscriptions. Enterprise distribution through three hyperscalers plus Oracle widens the funnel.

What Google gets from hosting a rival

Google Cloud has reasons of its own. Its annualized revenue growth accelerated to about 82% in the second quarter from 63% in the first, according to CNBC, while AWS grew 37% and Microsoft's Azure 42%.[3] Carrying OpenAI's models strengthens the case that Google Cloud is a neutral place to run AI, rather than a storefront for Gemini.

"We are focused on giving customers a broad selection of models, including those from OpenAI, so they can choose the right model for any given job, helping to deliver optimal quality and lower costs," said Thomas Kurian, CEO of Google Cloud.[2]

The subtext is competition with Gemini 4 Argon, Google's own flagship released in September.[3] Hosting a rival's best models looks generous, but it also keeps customers inside Google's billing and governance perimeter when they decide Gemini is not the answer for a particular job.

What developers get on day one

Access starts in private preview for select enterprise customers. General availability is planned "in the coming weeks," according to OpenAI.[2] Teams can call the models through Google Cloud APIs, and local Codex CLI and Desktop workflows can connect to the Gemini platform for inference.[1]

OpenAI models will also appear inside Google Antigravity and the Gemini Agent, though Google lists both as coming soon.[1][2] That matters for coding agents, where model choice and tool integration often decide whether a workflow is practical.

Google is promising strict data perimeters. Prompts and outputs stay inside the customer's designated boundary and are not used to train foundation models. The models run on dedicated infrastructure behind isolated gateways, with Zero Data Retention and Confidential Compute listed as coming soon, alongside VPC Service Controls and unified audit logging.[1]

"Google Cloud is a leading platform for developers and businesses building AI agents, writing software, and putting AI to work," said Greg Brockman, OpenAI's president and co-founder. "This partnership is an important step in making our most capable models and Codex available where customers are already building."[2]

The fine print

The announcement leaves several questions open. Neither company published prices, rate limits or a firm GA date. The private preview is limited to select enterprise customers, and the sign-up form asks interested teams to request access.[1][2]

Microsoft remains OpenAI's primary cloud partner, OpenAI said in its post, even as models spread to AWS, Oracle and now Google.[2] Regulated industries are the explicit target. The pitch rests on governance and procurement convenience as much as raw model quality.[2]

For now the practical effect is narrower than the headlines suggest. A developer outside the preview cannot yet buy GPT-6 tokens on Google Cloud. But the direction is clear. OpenAI is turning its models into something enterprises can buy wherever they already run infrastructure, and the cloud providers are competing to be the checkout counter.

Sources

1. Google Cloud blog, "Google Cloud expands model choice and flexibility, bringing OpenAI models to customers," Oct. 21, 2026
2. OpenAI, "OpenAI models are coming to Google Cloud," Oct. 21, 2026
3. CNBC, "OpenAI to bring AI models to Google's cloud, in further expansion beyond Microsoft," Oct. 21, 2026
4. OpenAI, "Access OpenAI models and Codex through your Oracle cloud commitment," June 10, 2026, updated Aug. 7, 2026