Google is now selling its biggest rival’s AI models. On October 21, Google Cloud announced that OpenAI’s GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna are coming to its Model Garden, the catalog of more than 200 models enterprises can buy through Google Cloud. The rollout starts in private preview for select enterprise customers, with general availability promised in the coming weeks.
It is a strange sentence to type. Google builds Gemini, one of the three leading closed AI model families; OpenAI builds another. Until this week, buying OpenAI’s models through Google’s cloud was not an option. Now Google says it is the only major cloud offering all three, Gemini, Anthropic’s Claude and OpenAI’s GPT, on one platform.
The Deal Nobody Expected
The lineup covers the full spread of OpenAI’s current stack. GPT-6 Astra is the flagship, aimed at demanding coding, computer use, professional work and scientific research. GPT-6.1 Sol offers near-Astra intelligence at a lower cost for agentic coding and professional tasks. GPT-6 Luna is the fast, cheap option for everyday work at scale.
Developers in the preview can call the models through Google Cloud’s APIs; support inside Antigravity, Google’s coding tool, and Gemini for business is coming soon.
One Invoice to Rule Them All
Here is the part of the announcement that actually changes buying behavior. OpenAI models are transacted through Google Cloud Marketplace, so usage lands on a single Google Cloud invoice, draws down existing cloud spending commitments, and counts toward Flexible Savings Plans across model providers. There is no separate vendor contract to negotiate and no separate AI budget to carve out.
That matters because enterprise cloud deals run on commitments: companies pre-pay for cloud spending in exchange for discounts, then look for ways to burn it down. Until now, OpenAI spend sat outside that pile for Google Cloud customers. Now it counts against it.
The Six-Month Sprint to Every Cloud
The Google deal completes a remarkably fast tour. In April, OpenAI’s exclusive arrangement with Microsoft loosened and Amazon began hosting its models. Oracle followed in June, selling OpenAI models against Oracle cloud commitments. Google is now the fourth channel, and OpenAI models run on all three of the big US clouds.
Microsoft, OpenAI was careful to add, remains its primary cloud partner. What has changed is the shape of the business: OpenAI is no longer a single-cloud company, and the clouds are becoming model supermarkets, competing on breadth of shelf, billing and governance rather than on one house model alone.
The Fine Print Buyers Should Read
Three caveats deserve attention before anyone rewrites their AI strategy. First, this is a private preview, not a public launch. Neither company has published a general availability date beyond “the coming weeks,” and eligibility is limited to select enterprise customers.
Second, Google has not published its own prices for the OpenAI models. SiliconANGLE reports OpenAI’s own API list prices, with Sol costing roughly a fifth of Astra and Luna roughly a hundredth, but Google Cloud pricing may differ, so do not assume the API price list carries over.
Third, the strongest data protections are still on the way. Google says prompts, context and outputs stay inside the customer’s data perimeter and are never used to train foundation models, but Zero Data Retention and Confidential Compute options are listed as coming soon. Teams with sensitive workloads should confirm what is live before migrating anything regulated.
Why This Matters
The commodity version of this story is “rivals become partners,” and it misses the point. The real shift is that distribution, not model quality, has become the scarce asset in enterprise AI. Gemini, Claude and GPT are close enough on many workloads that procurement convenience, billing and governance now decide deals, and Google just monetized that reality by renting shelf space to its rival.
For AI teams, the smart move is practical: run a small pilot on real tasks, benchmark GPT, Claude and Gemini side by side, and build a model gateway that lets you switch without rewriting your app. Google’s announcement makes that flexibility cheaper to buy. It does not make the choice of model any less important.
Meanwhile in AI Tech: OpenAI’s GPT-6.1 Sol pricing, explained and Google’s Gemini 4 Argon launch at half price show how the pricing war got here.
