OpenAI and Google Cloud announced on October 21, 2026 that OpenAI's frontier models, along with its Codex agentic workflows, will be available in Google Cloud's model garden. Both companies published their own announcements the same day: OpenAI framed the deal as expanding where enterprises can use its models, while Google Cloud framed it as completing its model lineup - the company says it is now the only hyperscale cloud provider offering access to all three leading closed-source LLMs, alongside more than 200 first-party, third-party and open models on its platform.
The move puts OpenAI's models in front of Google Cloud's enterprise base for the first time, and it chips at the assumption that OpenAI workloads live on Microsoft Azure. OpenAI's own announcement is careful on that point: it states explicitly that Microsoft remains its primary cloud partner. What changes is that Azure is no longer the only hyperscale option for customers who want OpenAI's frontier models inside a managed enterprise cloud.
What is actually launching
Google Cloud's post names the three models entering model garden. GPT-6 Astra is positioned for demanding work across coding, computer use, professional tasks and scientific research. GPT-6.1 Sol offers what Google describes as near-Astra intelligence at lower cost, aimed at agentic coding, computer use and professional work. GPT-6 Luna is the fast, cost-efficient model for everyday work at scale. The models are rolling out now in private preview to select enterprise customers, with general availability planned in the coming weeks, according to OpenAI.
Beyond raw model access, the integration covers OpenAI's Codex agentic workflows: starting today, Google says customers can connect their local Codex CLI and Desktop workflows directly to Google's Gemini platform for inference. OpenAI models will also come to Google Antigravity and the Gemini Agent - and, on Google's side, to Gemini for business - giving enterprise users model choice inside Google's own agentic surfaces rather than only through the API.
The commercial plumbing matters
The purchasing mechanics explain why this deal could move real workloads. OpenAI models in model garden are transacted through Google Cloud Marketplace, so usage lands on a single Google Cloud invoice. Enterprises with Google Cloud spend commitments can draw down those commitments for these purchases, and Google says usage counts toward its Flexible Savings Plans across model providers. For a large customer already committed to GCP spend, that turns adopting OpenAI models from a new vendor negotiation into an accounting line - and it removes the separate-contract friction that has historically kept enterprises inside one model ecosystem.
Governance is the enterprise pitch
Both announcements lead heavily on data governance, the usual blocker for regulated industries. OpenAI's post says the integration pairs its models with Google Cloud's built-in access controls, centralized data governance and high-performance infrastructure for demanding enterprise workloads. Google goes further into mechanics: customer prompts, context and outputs remain entirely within the customer's designated data perimeter and are never used to train foundation models, with Zero Data Retention options listed as coming soon. OpenAI models run on dedicated Google Cloud infrastructure, queries are routed through hardened, isolated gateways, and Confidential Compute is also listed as coming soon. Every agent and model interaction operates under an authenticated identity, with VPC Service Controls protection and unified audit logging.
Box is the named enterprise voice in Google's post: Yashodha Bhavnani, VP of AI Products at Box, says that by using OpenAI models through Google Cloud, the company will be able to pair OpenAI's frontier models with Google's security and compliance capabilities for enterprise content. It is a customer endorsement Google clearly wanted on record at launch.
The executives' framing
Greg Brockman, OpenAI's president and co-founder, called Google Cloud a leading platform for developers and businesses building AI agents, and said the partnership makes OpenAI's most capable models and Codex available where customers are already building. Thomas Kurian, Google Cloud's CEO, kept his quote to model economics: a broad selection of models, including OpenAI's, so customers can choose the right model for any given job at optimal quality and lower cost.
Read together, the two posts describe a deliberate swap of exclusivity for reach. OpenAI gets distribution inside a rival's enterprise estate without giving up its Microsoft relationship, which it named explicitly in its own post. Google Cloud gets the missing member of its model roster and a fresh argument that its model garden is the neutral place to run mixed-model workloads. Enterprises get something simpler: the ability to route each task to the model that fits it - including, now, OpenAI's - without leaving their existing Google Cloud contracts. For developers, the practical takeaway is nearer-term: if your organization already runs on Google Cloud, the request-access form for the private preview is live on both companies' posts, and Codex CLI and Desktop can point at Google's platform for inference today. The open questions are the ones neither post answers: exact pricing per model, the general-availability date beyond "coming weeks", how the preview's enterprise controls evolve into their "coming soon" Zero Data Retention and Confidential Compute tiers, and how Microsoft responds to its partner's models arriving, by invitation, on a competitor's cloud.