OpenAI's models are coming to Google Cloud. Microsoft is still the primary partner.

OpenAI and Google Cloud announced on October 21 that OpenAI's frontier models and its Codex coding workflows are coming to Google Cloud's model garden. With Amazon Web Services already hosting OpenAI models since April, that puts OpenAI on all three of the largest US clouds.

What is live today. Google Cloud says GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna are "rolling out now in private preview to select enterprise customers," with an interest form for everyone else. Customers in the preview can call the models through Google Cloud's APIs and point their local Codex CLI and Codex desktop app at Google's Gemini platform for inference. OpenAI says general availability is "planned in the coming weeks."

How you pay. The models are sold through Google Cloud Marketplace. Google says usage appears on a single Google Cloud invoice, draws down existing Google Cloud spend commitments and counts toward its Flexible Savings Plans. For a company that has already committed a budget to Google Cloud, that removes the need for a separate OpenAI contract. Neither company published per-token prices, and when we checked Google's OpenAI model docs and generative AI pricing page on the morning of October 21, they still listed only the open-weight gpt-oss models.

What Google promises on data. Prompts, context and outputs stay inside the customer's data perimeter and are "never used to train foundation models," per Google. The models run on dedicated Google Cloud infrastructure, behind isolated gateways, under VPC Service Controls and unified audit logging. Two features are marked "coming soon": zero data retention options and Confidential Compute.

What's next. Both companies say OpenAI models will "soon" appear in Google Antigravity, Google's coding tool, and in Gemini Agent and Gemini for business. No date was given.

Where Microsoft stands. OpenAI's post ends with one line: "Microsoft remains OpenAI's primary cloud partner." Reuters reported in April that the two renegotiated their deal so Microsoft no longer had the exclusive right to sell OpenAI's models, which cleared the way for Amazon and Google. CNBC notes that Google, Amazon and Microsoft all report quarterly results next week.

Google's claim. Google calls itself "the only hyperscale cloud provider offering access to all three leading closed source LLMs." It doesn't name the three. Our reading: Google's own Gemini, OpenAI's GPT models and Anthropic's Claude, which Google Cloud already offers.

Our take

For most developers, nothing changes today: the preview is invite-only and there is no price list. For teams whose AI budget lives inside a Google Cloud commitment, it is a real change, because they can soon use GPT models without a new vendor contract or a second bill. Before switching any production traffic, wait for the published per-token prices and for zero data retention to actually ship.

What we did and didn't do: everything above comes from OpenAI's and Google Cloud's announcements, Google Cloud's documentation and pricing pages, read on October 21, 2026, plus reporting by CNBC and Reuters. MODEL has not used the preview. This is not sponsored and has no affiliate links. Not investment advice.

MODEL Editors