OpenAI Models and Codex Are Coming to Google Cloud in a Surprise Rival Deal

OpenAI models on Google Cloud are no longer a hypothetical. On October 21, OpenAI announced that it is partnering with Google Cloud to bring its frontier models and Codex agentic workflows into Google Cloud's Model Garden, the catalog where businesses already pick and deploy AI models. For two companies that compete head to head on chatbots, search and coding tools, it is an unusual handshake, and it says a lot about where the money in AI is moving: toward the cloud contracts that big companies have already signed.

What OpenAI and Google Cloud announced

According to OpenAI's announcement, Google Cloud customers will be able to access OpenAI's frontier models and Codex workflows natively inside Model Garden. OpenAI describes the setup as being backed by Google Cloud's data governance and infrastructure, which matters to companies that have spent years building security and compliance rules around one cloud provider.

The buying process is the part enterprise customers will care about most. OpenAI says the models can be purchased through Model Garden and transacted through the Google Cloud Marketplace. Companies that already have Google Cloud spend commitments can draw down those commitments for these purchases. AICoder, a developer news site that covered the deal, notes this means customers would not need a separate OpenAI contract or invoice.

OpenAI also said its models will soon be available in Google Antigravity and the Gemini Agent, two of Google's own agent products. That detail stands out. It suggests Google is willing to let a rival's models run inside tools that carry the Gemini name.

General availability is planned "in the coming weeks," according to OpenAI. For now, interested companies can request access through a form linked in the announcement.

Which models are included

OpenAI's post does not list specific model names. AICoder reports, citing Reuters, that GPT-6 Astra is part of the offering. Until Google Cloud publishes its own model list and pricing, buyers should treat the exact lineup as unconfirmed. If you are tracking OpenAI's recent product moves, our coverage of the GPT-6 intelligent UI rollout in ChatGPT gives useful background on the newest models.

Why OpenAI models on Google Cloud matter

The simplest way to understand this deal is to follow the budget. Large companies rarely buy AI the way a hobbyist does, with a credit card and an API key. They sign multiyear cloud agreements worth millions of dollars and commit to spending a set amount. Any product that can be bought against that commitment becomes far easier to approve, because the money is already set aside.

By putting its models in Google Cloud Marketplace, OpenAI gets access to that pool of committed spending. Google Cloud, in turn, gets to tell customers that they do not have to leave its platform to use OpenAI's tools. Thomas Kurian, chief executive of Google Cloud, framed it around choice, saying in OpenAI's announcement that customers want "a broad selection of models, including those from OpenAI."

Greg Brockman, OpenAI's president and co-founder, made a similar point from the other side. He said the partnership is about "making our most capable models and Codex available where customers are already building."

The Microsoft question

Microsoft has been OpenAI's closest partner for years, so any deal with a rival cloud draws attention. OpenAI addressed it directly in one line of its announcement: "Microsoft remains OpenAI's primary cloud partner."

That sentence does a lot of work. It signals that OpenAI is widening its reach without, at least officially, stepping away from Microsoft Azure. AICoder points out that OpenAI models are already offered on Amazon Web Services and says this deal puts them on all three major US clouds. OpenAI's own site also lists a June 2026 post about accessing its models and Codex through Oracle cloud commitments, which shows the same strategy playing out across several providers.

Who is affected

Enterprise IT and procurement teams

Companies that standardized on Google Cloud have often faced a choice: use Google's Gemini models inside their existing setup, or open a separate relationship with OpenAI. This deal removes much of that friction. Procurement teams can buy through a marketplace they already use, and security teams can keep data inside a cloud environment they already audit.

Developers using Codex

Codex is OpenAI's agentic coding tool. Bringing Codex workflows into Model Garden means developers working inside Google Cloud could use OpenAI's coding agent without switching platforms. For teams comparing coding assistants, that makes side by side testing easier. It also raises the stakes for Google's own coding tools, which now share shelf space with a direct competitor.

Google's Gemini business

Google is betting that being the place where customers run many models is worth more than keeping its own models exclusive. That is a familiar cloud strategy, but it carries risk. If customers inside Google Cloud choose OpenAI over Gemini, Google still earns cloud revenue, yet its own model business may feel the squeeze. Recent changes to Gemini's consumer tiers, such as the shift that left free Gemini users with Flash Lite only, show Google is already juggling cost and demand across its lineup.

What businesses should do now

If your company runs on Google Cloud and is curious about OpenAI's models, a few practical steps make sense before general availability arrives.

1. Check your commitment terms. Ask your Google Cloud account team whether Marketplace purchases of OpenAI models will count fully toward your existing commitment, and whether any limits apply.
2. Wait for pricing. Google Cloud has not published pricing for these models yet. Compare it against OpenAI's direct API rates once both are available.
3. Review data handling. Confirm where prompts and outputs are processed and stored, and how that fits your compliance rules.
4. Plan a small pilot. Pick one workload, such as a coding task for Codex or a document summary job, and test OpenAI models against the Gemini models you already use.

Spending discipline matters here. AI infrastructure costs have drawn more scrutiny lately, as seen in the pressure on CoreWeave's stock as bond yields rose. Finance teams will want hard numbers before approving broad rollouts.

What to watch next

The first thing to watch is the general availability date and the final model list. OpenAI has only said "the coming weeks," and Google Cloud has not posted pricing. Those two details will decide how attractive the offer really is.

The second is how deeply OpenAI's models show up in Google Antigravity and the Gemini Agent. A rival model sitting inside a product branded Gemini would be a notable shift in how Google presents its AI.

The third is how Microsoft responds. OpenAI's statement about Microsoft being its primary cloud partner is clear, but competition among cloud providers for AI spending is only getting sharper. OpenAI has also been expanding in other directions, including its move to acquire Orchestra Bio, so its strategy is clearly about reach on many fronts. For more coverage like this, visit our AI news section.

Frequently asked questions

When will OpenAI models be available on Google Cloud?

OpenAI says general availability is planned in the coming weeks. Companies can request access now through a form linked in OpenAI's October 21 announcement.

Can I use my existing Google Cloud budget to pay for OpenAI models?

Yes, according to OpenAI. Enterprises with Google Cloud spend commitments can draw them down for purchases of OpenAI models made through the Google Cloud Marketplace.

Is OpenAI leaving Microsoft Azure?

No. OpenAI stated in its announcement that Microsoft remains its primary cloud partner. The Google Cloud deal adds another channel rather than replacing Azure.

Does this include Codex?

Yes. OpenAI says Codex agentic workflows are part of what is coming to Google Cloud's Model Garden, alongside its frontier models.