OpenAI Models Come to Google Cloud Model Garden: Private Preview, Billing and What Builders Should Check

OpenAI and Google Cloud announced on October 21 that OpenAI's frontier models and Codex workflows are coming to Google Cloud's Model Garden. Today it is a private preview for select enterprise customers, and general availability is described as "planned in the coming weeks." If you build on Google Cloud, the practical changes are about billing, data controls and model choice, and several of the controls are labelled "coming soon."

I read both primary posts: OpenAI's OpenAI models are coming to Google Cloud and Google Cloud's Google Cloud brings OpenAI models to customers. I have not tried the preview, seen pricing or opened the access forms.

What is announced

According to Google Cloud's post, three models are rolling out now in private preview to select enterprise customers:

• GPT-6 Astra, described as built for demanding work across coding, computer use, professional tasks and scientific research.
• GPT-6.1 Sol, described as near-Astra intelligence at lower cost across agentic coding, computer use and professional work.
• GPT-6 Luna, described as a fast, cost-efficient model for everyday work at scale.

Google also says that starting today customers can build with these models through its APIs and connect local Codex CLI and Desktop workflows to its Gemini platform for inference. The OpenAI post says the integration brings frontier models and Codex agentic workflows into Model Garden.

Both sides say OpenAI models will soon come to Google Antigravity. Google's post also names Gemini for business, and OpenAI's names the Gemini Agent. Neither gives a date.

Billing and purchasing

Per the OpenAI post, customers will buy the models through Model Garden, transacted through the Google Cloud Marketplace, and enterprises with Google Cloud spend commitments can draw down those commitments. Google adds that usage appears on a single Google Cloud invoice and counts toward Flexible Savings Plans across model providers.

Neither post I read lists prices. I would not assume Google Cloud pricing matches OpenAI's own API pricing.

Data and security claims

Google's post says:

• Prompts, context and outputs stay within the customer's data perimeter and are not used to train foundation models.
• Zero Data Retention options are "coming soon."
• OpenAI models run on dedicated Google Cloud infrastructure, with Confidential Compute "coming soon."
• Agent and model interactions run under an authenticated identity, with VPC Service Controls and unified audit logging.

The OpenAI post stresses access controls and centralized data governance for regulated industries. Both are vendor statements. Note that ZDR and Confidential Compute are not yet available, per Google's own wording.

Context

Google calls itself the only hyperscale cloud provider offering access to all three leading closed-source LLMs. That is a Google marketing claim, and I did not verify it. OpenAI's post says Microsoft remains its primary cloud partner.

What I would check

These are my own untested suggestions.

1. Are you in the preview? Access is limited to select enterprise customers. Use the interest form each post links to.

2. Model IDs and API shape. Check whether the calls match the OpenAI API you already use or a Google Cloud one.

3. Region and data residency. The posts do not list regions in the parts I read.

4. Pricing and quotas. Ask for them before you plan a migration, since neither post states them.

5. Which security features are live. ZDR and Confidential Compute are described as upcoming.

6. Codex setup. Test the Codex CLI connection in a non-production project first.

7. Contract terms. If you buy through Marketplace, check how commitment drawdown and support work for these models.

What I would not conclude

• That the models are generally available. The posts say private preview, with GA planned in the coming weeks.
• That prices, limits or features match OpenAI's own platform.
• That the data protections described are all active today.

Frequently asked questions

Which OpenAI models are coming to Google Cloud?

GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna, per Google Cloud's post. It says more are coming soon.

Is it generally available?

No. Google says private preview for select enterprise customers. OpenAI says general availability is planned in the coming weeks.

How do I pay for it?

Through Google Cloud Marketplace, on a Google Cloud invoice, and against existing Google Cloud commitments, per both posts.

Can I use Codex with it?

Google says you can connect local Codex CLI and Desktop workflows to its Gemini platform for inference starting today. I did not test this.

Are my prompts used for training?

Google says customer prompts, context and outputs are never used to train foundation models. This is Google's statement.

Is Microsoft still OpenAI's main cloud partner?

OpenAI's post says Microsoft remains its primary cloud partner.

What is still missing?

Public pricing, GA dates, ZDR and Confidential Compute are not available in the posts I read. Google Antigravity availability is "coming soon."

Sources

• OpenAI, OpenAI models are coming to Google Cloud (October 21, 2026).
• Google Cloud Blog, Google Cloud brings OpenAI models to customers (October 21, 2026).