The OpenAI models Google Cloud partnership, announced on 21 October 2026, brings OpenAI’s frontier models and Codex agentic workflows into Google Cloud’s Model Garden. If you build on Google Cloud, you will soon be able to buy and run OpenAI’s newest models where your data and workloads already live, pay for them through your existing Google Cloud commitments, and connect Codex to Google’s infrastructure for inference. Here is what is actually available, how it works, and what to do while you wait for general availability.
What OpenAI is bringing to Google Cloud
OpenAI and Google Cloud announced the partnership on 21 October 2026, in matching posts from both companies. OpenAI’s frontier models and Codex agentic workflows are coming into Google Cloud’s Model Garden, the catalogue where Google Cloud customers browse and deploy models. The point, in OpenAI’s words, is native access: customers get the models inside the environment where they already work, backed by Google Cloud’s data governance and high-performance infrastructure.
Google Cloud named three models rolling out now:
- GPT-6 Astra — designed for demanding work across coding, computer use, professional tasks, and scientific research.
- GPT-6.1 Sol — near-Astra intelligence at lower cost.
- GPT-6 Luna — a fast, cost-efficient model for everyday work at scale.
The models are in private preview for select enterprise customers. General availability is planned in the coming weeks, according to OpenAI, and customers can request access through an interest form now. Google Cloud says it is now the only hyperscale cloud provider offering access to all three leading closed-source large language models — Gemini, Claude, and GPT — alongside more than 200 first-party, third-party, and open models on a single enterprise AI platform.
How access and billing work
The purchase route is what changes most for Google Cloud customers. OpenAI models are bought through Google Cloud Marketplace, and the usage lands on a single Google Cloud invoice. Enterprises with existing Google Cloud spend commitments can draw those commitments down against the purchases, and usage counts toward Flexible Savings Plans across model providers.
That matters because it simplifies how teams buy AI: they transact through the cloud bill they already pay, and the catalogue puts multiple model families on one platform to choose from.
What Google has not published yet is its own pricing for the three models. Do not assume the prices match OpenAI’s direct API prices until your account team confirms them — and the same applies to preview quotas and regions, which the announcements do not specify.
Codex and agentic workflows on Google Cloud
The model list is only half the announcement. The other half is Codex: OpenAI’s agentic coding tool. Starting now, customers can connect their local Codex CLI and Desktop workflows directly to Google’s Gemini platform for inference, so agentic coding work runs against Google Cloud infrastructure.
That is the part aimed squarely at software teams. Google Cloud has long pitched itself as a destination for AI workloads and the developers building them, and bringing Codex into that environment lets teams modernize their software development lifecycle with agentic workflows.
“Google Cloud is a leading platform for developers and businesses building AI agents, writing software, and putting AI to work. This partnership is an important step in making our most capable models and Codex available where customers are already building, giving them even more choice in how they use frontier AI as we grow the agentic ecosystem together.”
— Greg Brockman, President and co-founder, OpenAI
Model choice: OpenAI models in Antigravity and the Gemini Agent
OpenAI models are also coming soon to Google Antigravity and the Gemini Agent, Google’s business-agent product. The idea is model choice inside the tools developers and business users already open every day, rather than a separate destination for each model family.
The pitch from Google Cloud is that enterprise AI is about matching each workload to the right model, not using one model for everything. With OpenAI’s models in the catalogue, a team could use Astra for a demanding coding task, Sol for routine agentic work at lower cost, and Luna for high-volume everyday work — and switch between them as the job changes.
“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.”
— Thomas Kurian, Chief Executive Officer, Google Cloud
What regulated industries get from the integration
For organizations in regulated industries, the announcement leans heavily on governance. Deploying AI in those settings requires control over who can access sensitive data and how it is used, and the integration pairs OpenAI’s models with Google Cloud’s built-in access controls, centralized data governance, and high-performance infrastructure.
What the announcements confirm is that customer prompts, context, and outputs remain entirely within the customer’s designated data perimeter and are never used to train foundation models. The integration pairs that with Google Cloud’s access controls and centralized data governance.
Coming soon, by Google’s own labelling: Zero Data Retention options and Confidential Compute. Treat those as roadmap items, not features you can rely on today — the announcements are explicit that they are not yet available.
“By using OpenAI models through Google Cloud, we will be able to give customers access to OpenAI’s frontier AI models alongside Google Cloud’s security and compliance capabilities, enabling them to get more value from their enterprise content.”
— Yashodha Bhavnani, VP of AI Products at Box
Where this fits in the cloud AI landscape
The partnership is striking partly because OpenAI and Google compete head-on in AI products and search. It makes sense once you see the distribution picture. Reuters reports that OpenAI’s exclusive distribution arrangement with Microsoft expired in April 2026, and OpenAI has since expanded distribution to Amazon Web Services and now to Google Cloud. OpenAI has also offered access through Oracle Cloud. Microsoft, for its part, remains OpenAI’s primary cloud partner, as OpenAI’s own announcement states.
For Google Cloud, the deal strengthens a strategy of being a neutral platform with the widest choice of models, rather than relying only on its own Gemini family. For OpenAI, it is distribution: selling frontier models to more enterprise customers through the clouds those customers already buy from. The two companies are rivals in products and partners in distribution — and that tension is the story of this deal as much as the technology.
OpenAI models Google Cloud: what to do now
If you build on Google Cloud, the practical steps are straightforward. First, request access through the interest form if you are not already in the private preview — preview places are going to select enterprise customers first. Second, evaluate the three models against your actual workloads: Astra for the hardest coding and research work, Sol where near-Astra quality at lower cost is enough, Luna for everyday work at scale. Third, talk to your Google Cloud account team about pricing, regions, and quotas before planning a production rollout, because none of those are published yet.
What is still unknown is worth stating plainly: the exact general availability date, Google Cloud’s own prices for the three models, which regions the preview covers, and when Zero Data Retention and Confidential Compute will actually ship. The announcements give you the shape of the deal and the buying mechanics. The details that decide a production deployment are still to come — and the path OpenAI sketches, from evaluation to broader adoption, starts with getting into the preview and testing the models on your own workloads.
