OpenAI Partners With Google Cloud to Bring Frontier Models and Codex to Model Garden
OpenAI and Google Cloud have announced a partnership that will bring OpenAI frontier models and Codex agentic coding workflows to Google Cloud customers through Model Garden.
The move gives enterprises another way to use OpenAI technology while keeping development, purchasing, governance, and deployment within their existing Google Cloud environment. It also marks an important expansion of OpenAI’s multi-cloud strategy.
What the partnership includes
Google Cloud customers will be able to access OpenAI models natively through Model Garden. OpenAI says organizations will be able to use Google Cloud’s infrastructure, access controls, and governance tools while building with those models.
Purchasing will be available through Google Cloud Marketplace. For enterprises that already have committed Google Cloud spending, eligible OpenAI purchases can count toward those commitments. That detail could make adoption substantially easier for large organizations whose procurement and security processes are already centered on Google Cloud.
Codex is also part of the partnership. The companies plan to support Codex through its app, command-line interface, IDE integrations, and API. OpenAI says its models are also expected to appear in Google Antigravity and Gemini Agent, giving developers additional ways to use them inside Google’s ecosystem.
Availability starts with a limited preview
This is not a full public launch yet. OpenAI is accepting requests for a limited preview that includes access to frontier models through the Responses API on Google Cloud and access to Codex workflows.
According to the preview terms, model access is intended for approved development and internal testing rather than production use. OpenAI says general availability is planned in the coming weeks, so production teams should wait for the final service terms, regional availability, supported-model list, pricing, quotas, and reliability commitments before making deployment decisions.
Organizations interested in early access can request the Google Cloud preview from OpenAI.
Why this matters for enterprise AI teams
For enterprise buyers, the most important part of the announcement is not simply that another model catalog is getting OpenAI models. It is that teams may be able to use OpenAI while staying inside an established Google Cloud control plane.
- Procurement becomes simpler. Marketplace purchasing can reduce the need to establish a separate vendor path.
- Governance stays familiar. Teams can use existing Google Cloud access and oversight practices around AI workloads.
- Developers gain more choice. OpenAI models and Codex can become options alongside the other models already available through Model Garden.
- Committed cloud spend may go further. Eligible purchases can be applied to existing Google Cloud commitments.
The partnership could be especially useful for companies that have standardized on Google Cloud but want to evaluate OpenAI models without building a separate cloud operating model.
Microsoft is still OpenAI’s primary cloud partner
The Google Cloud agreement does not end OpenAI’s relationship with Microsoft. OpenAI explicitly says Microsoft remains its primary cloud partner.
Instead, the announcement fits OpenAI’s broader infrastructure strategy. The company has described a growing cloud portfolio that includes Microsoft, Oracle, AWS, CoreWeave, and Google Cloud. The goal appears to be more computing capacity and more ways for customers to access OpenAI products in the environments they already use.
What to watch next
The announcement establishes the direction, but several practical details still matter. Enterprise teams should watch for the exact model catalog, regions, pricing, data-handling terms, rate limits, service-level commitments, and the production launch date.
It will also be worth watching how OpenAI models are presented next to Google’s Gemini family inside Model Garden, and how Codex integrates with Google’s developer tools. If the implementation is smooth, the partnership could make model choice less dependent on cloud choice—and make enterprise AI stacks more competitive as a result.
Source: OpenAI’s official partnership announcement, published October 21, 2026.
