The OpenAI and Google Cloud logos. Image: OpenAI and Google Cloud logos via Simple Icons (CC0)
OpenAI’s top AI models are about to go on sale inside the cloud of the company behind Gemini, its biggest rival. OpenAI and Google Cloud said on Wednesday that GPT-6 Astra and other OpenAI frontier models, along with its Codex coding agent, are coming to Model Garden, Google Cloud’s catalogue of more than 200 AI models for businesses.
The deal means OpenAI’s models will be available on all three of the biggest cloud platforms: Microsoft Azure, Amazon Web Services and now Google Cloud. For years Microsoft had them to itself.
Which OpenAI models are coming to Google Cloud
Google’s own announcement names three models, “rolling out now in private preview to select enterprise customers”, with more to follow:
- GPT-6 Astra: OpenAI’s flagship, which Google describes as built for “demanding work across coding, computer use, professional tasks, and scientific research”.
- GPT-6.1 Sol: “near-Astra intelligence at lower cost” for agentic coding, computer use and office work.
- GPT-6 Luna: the fast, cheap model for everyday work at scale.
Google says customers in the preview can already call the models through its APIs and point their local Codex command-line tool and Codex desktop app at Google’s Gemini platform for inference. OpenAI’s post says general availability is planned “in the coming weeks”. Companies that want in now have to fill in an interest form on either company’s site.
Neither company has said which countries or cloud regions will get the models first, or what they will cost through Google compared with OpenAI’s own API.
Paid for out of existing Google Cloud budgets
The selling point for big customers is the billing. The models are bought through Google Cloud Marketplace, so usage lands on a single Google Cloud invoice and counts against the spending commitments many large companies have already signed with Google. Google says it also counts towards its Flexible Savings Plans, which cover models from several providers.
Google is pitching the data side to banks, healthcare firms and other regulated businesses. It says customers’ prompts and outputs stay inside their chosen data perimeter and are never used to train models, that the OpenAI models run on dedicated Google Cloud infrastructure, and that requests go through isolated gateways so model providers can’t reach a customer’s cloud environment. Zero data retention and Confidential Computing are both listed as “coming soon”, so neither is there at launch.
Coming to Antigravity and the Gemini agent
The two companies say OpenAI’s models will also arrive “soon” in Google Antigravity, Google’s AI coding environment, and in the Gemini agent, the workplace assistant Google Cloud launched this month, which already lets companies run it on Anthropic’s Claude models as well as Gemini. No date was given for either.
That leaves Google in an unusual position: selling OpenAI’s best model next to its own Gemini 4 Argon. Google says it is now the only one of the big cloud providers to offer “all three leading closed source LLMs”, meaning Gemini, Claude and GPT, on one platform. Anthropic is one of Google Cloud’s biggest customers: its draft IPO filing lists about $111 billion of computing agreements with Google Cloud.
Microsoft is no longer the only door
OpenAI’s announcement ends with a single line for Redmond: “Microsoft remains OpenAI’s primary cloud partner.” But the exclusive arrangement is gone. Microsoft’s exclusive licence to OpenAI’s technology ended in April, and the next day Amazon said it would host OpenAI models, CNBC reported. At DevDay last month OpenAI added Bedrock Managed Agents, built with Amazon so companies can run OpenAI agents entirely inside AWS.
The two companies are not new partners either: OpenAI said last year that it was using Google Cloud to help run ChatGPT and its API. What changes now is that Google becomes a shop window for OpenAI’s models, not just a landlord. Alphabet, Amazon and Microsoft all report quarterly results next week, so expect cloud executives to be asked about it.
Why it matters
Most big companies already have a cloud contract, and many of them are with Google; letting them buy GPT-6 Astra with money they have already committed removes one of the biggest obstacles to switching models. It is a bet by OpenAI that distribution matters more than loyalty, and a bet by Google that it earns more by hosting a rival than by shutting it out.
Sources: OpenAI, Google Cloud, OpenAI Developer Community, CNBC.
Latest OpenAI news
