In short: On 22 October 2026, Google Cloud announced that OpenAI's current models — GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna — are coming to its Model Garden in a private enterprise preview, according to TechWire Asia. In the same week, Google Cloud also confirmed that Anthropic can now run Claude directly for customers who want new features first, as WebProNews reported — making Google, in its own words, the only hyperscale cloud offering all three leading closed-source LLMs: Gemini, Claude and GPT.
What Google Cloud announced on 22 October
According to TechWire Asia's reporting dated 22 October 2026, Google Cloud revealed that OpenAI's current model lineup — GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna — will become available in Google Cloud's Model Garden as a private enterprise preview. The commercial mechanics are straightforward: customers purchase the OpenAI models through the Google Cloud Marketplace, and that spend counts toward their existing Google Cloud commitments.
Google's pitch to cautious enterprise buyers rests on two reassurances reported by TechWire Asia. First, prompts and model outputs stay inside the customer's data perimeter. Second, that data is not used for training. Google Cloud CEO Thomas Kurian summed up the strategy in a statement: "We are focused on giving customers a broad selection of models, including those from OpenAI."
The announcement also confirmed that OpenAI models in Gemini for business and in Google's Antigravity coding tool are "coming soon" — no firm dates attached.
The Anthropic piece: Claude runs directly on Google Cloud
Separately, WebProNews reported on 22 October that Google Cloud now lets Anthropic run Claude directly for customers who want new features first. The Claude lineup available through this arrangement includes Claude Fable 5.1, Sonnet 5.5, Haiku 5.5 and Opus 5.5.
Google's claim, per WebProNews: it is now the only hyperscale cloud provider offering all three leading closed-source LLMs — Gemini, Claude and GPT — on a single platform.
Neither the OpenAI nor the Anthropic announcement specified financial terms or pricing. The money side of both arrangements remains undisclosed.
Why this matters
For years, OpenAI's enterprise story was inseparable from Microsoft's cloud. This announcement gives OpenAI direct reach into enterprises already committed to Google Cloud — companies that can now buy OpenAI models through a marketplace contract they already have, with spend folded into their existing Google Cloud commitments.
For Google Cloud, the positioning is clear: stop competing purely model-on-model with Gemini and become the neutral shelf where enterprises shop for all three frontier families. Whether "offering all three" translates into pricing power or just a crowded menu will depend on the numbers, and those haven't been published yet.
For the market, the cloud model race is now visibly three-way. Every hyperscaler is assembling a multi-model line-up; differentiation moves from "who has the best model" toward "who has the broadest selection with the least procurement friction."
Procurement friction is the quiet engine here. Large enterprises negotiate Google Cloud committed-use agreements that represent real money on the table; letting OpenAI model spend draw down those commitments removes a contracting headache that has historically kept multi-cloud buyers in lane. The Model Garden placement matters for the same reason: enterprises prefer to discover and govern models inside a platform they already audit, with access controls, logging and billing in one place. The reported data-perimeter and no-training reassurances are table stakes for that audience — no regulated enterprise adopts a model whose prompts train someone else's frontier model.
The three-way menu is the strategy
Considered together, the two 22 October announcements describe a single strategy: Google Cloud as the neutral shelf. One hyperscaler now carries its own Gemini family, Anthropic's Claude (with direct deployment for customers who want new features first) and OpenAI's GPT — the three leading closed-source LLM families — side by side. The differentiation pitch is selection plus procurement simplicity rather than model supremacy.
That carries an implicit admission. Google's own Gemini models have improved rapidly, but enterprise AI adoption is fragmented: teams standardize on different model families for different workloads, and forcing a single family means losing the workloads you don't win. Becoming the platform where all three families coexist converts a model-race problem into a platform advantage — assuming customers can compare and switch without renegotiating everything each time.
There are also limits to what was announced. The OpenAI availability is a private enterprise preview, not general availability, so most of the market will be watching from outside. And while the Anthropic arrangement is described as a managed deployment, the WebProNews reporting does not detail how feature parity, versioning or support responsibilities are split between Anthropic and Google. Those operational details determine whether "all three on one cloud" is a genuine integration or a billing relationship.
What to watch
Two open questions. First: pricing — what happens to the enterprise value case when the Model Garden preview opens beyond its private audience, and how Marketplace rates for GPT-6 compare with Azure's and the API's. Second: delivery — whether the "coming soon" OpenAI integrations in Gemini for business and the Antigravity coding tool actually land, and on what timetable.
A third, subtler question: whether Microsoft answers. OpenAI models on Google Cloud erode the exclusivity that made Azure the default enterprise route to GPT. Azure still has deep integration advantages, but the cloud AI race is clearly moving from exclusive partnerships toward open shelves — and the vendor that aggregates best may matter more than the vendor that invents best.
Related reading: our 22 October article on Anthropic's launch of Claude Duet, a proactive agent app.
Be the first to share a useful question or thought.