# OpenAI is coming to Antigravity: what changes with Google Cloud

> Google announces Astra, Sol and Luna on Cloud, with OpenAI coming to Antigravity. What you can access today and what remains unknown about plans and pricing.


Published: 2026-10-21. Original source: [Google Cloud · OpenAI models and access announcement](https://cloud.google.com/blog/products/ai-machine-learning/google-cloud-brings-openai-models-to-customers).

**Google and OpenAI announced on October 21 that OpenAI models are coming to Antigravity.** The agreement starts with a private preview on Google Cloud for selected enterprise customers; Antigravity is among the additions planned next.

If you opened this story to find out whether you can already choose Astra or Sol in Antigravity, that distinction matters: **the announcement confirms the future integration but does not give an activation date or access terms for personal subscriptions**. [OpenAI plans general availability on Google Cloud in the coming weeks](https://openai.com/index/openai-models-google-cloud/).

## Where can you use OpenAI with Google, and what comes next?

The agreement brings together models, Google Cloud infrastructure and tools such as Codex. Each part has a different availability status. Choose a tool to explore it:

### Google Cloud · Private preview

**Access for selected enterprise customers.** Astra, Sol and Luna are rolling out in Model Garden to selected enterprise customers. Access is by request; general availability is planned for the coming weeks.

Purchases go through Google Cloud Marketplace and can draw down existing Google Cloud spending commitments.

[Request preview access](https://openai.com/form/openai-on-google-cloud/)

### Codex · Private preview

**Your local workflows, with inference on Google Cloud.** Google says customers in this preview can connect their Codex CLI and Desktop workflows to its Gemini platform for inference.

Codex remains the local tool. The change is a connection to Google Cloud infrastructure and enterprise access.

[Check the scope with Google](https://cloud.google.com/blog/products/ai-machine-learning/google-cloud-brings-openai-models-to-customers)

### Antigravity · Coming soon

**Confirmed in the announcement; no exact date yet.** Google and OpenAI announce that OpenAI models will come to Antigravity. The announcement does not name the specific models that will appear in its picker or give an activation date.

It also does not specify Google AI Pro or Ultra access, quotas or pricing for this integration.

[Read the OpenAI announcement](https://openai.com/index/openai-models-google-cloud/)

## Astra, Sol and Luna: the models confirmed for Google Cloud

[Google names three models for the Model Garden preview](https://cloud.google.com/blog/products/ai-machine-learning/google-cloud-brings-openai-models-to-customers). Model Garden is the catalog where Google Cloud teams discover and use models from different providers.

| Model | How Google describes it in this announcement |
| --- | --- |
| **GPT-6 Astra** | Demanding coding, computer use, professional work and scientific research. |
| **GPT-6.1 Sol** | Capabilities close to Astra at a lower cost, aimed at agentic coding and professional work. |
| **GPT-6 Luna** | Fast responses and cost efficiency for everyday work at scale. |

These are the provider’s descriptions. The announcement **does not publish a new comparison of performance, latency or cost per task** across the three. It also does not say which models will appear first in Antigravity’s picker. The list above applies to the Google Cloud preview.

## What changes if you already use Codex?

Google says preview customers can connect **their local Codex CLI and Desktop workflows to the Gemini platform for inference**: the processing through which a model generates its responses. That gives those tools a connection through the customer’s Google Cloud enterprise environment.

The model you choose, the environment that runs it and the tool you work in can now fit together differently. For a team already operating on Google Cloud, the practical benefit is adding OpenAI models within that environment, with its access controls and billing.

## Will it be included in Google AI Pro or Ultra?

**The announcements do not specify which personal plans will receive this Antigravity integration.** They also give no quotas, exact date or instructions for enabling it with Google AI Pro or Ultra.

The access Google describes today is a **private preview for selected enterprise customers**. The [OpenAI access request form](https://openai.com/form/openai-on-google-cloud/) collects interest in that preview; submitting it does not establish that your account is enabled.

If you already use [Gemini 4 Argon in Antigravity](/en/guides/gemini-4-argon-antigravity-pro/), its access terms belong to a separate announcement. The documentation for this integration still needs to identify the supported models and plans before you can know what your subscription includes.

## How will model usage be billed?

Purchases will go through **Google Cloud Marketplace**. Google says usage will appear on a Google Cloud invoice and can draw down existing cloud spending commitments. It will also count toward Flexible Savings Plans across model providers.

The announcements do not list a price per million tokens for this integration. “Lower cost” explains how Google positions Sol relative to Astra, but it is not enough to calculate what your team will pay through this new channel.

## When can you try OpenAI in Antigravity?

For now, the official answer is **coming soon**. OpenAI places Google Cloud general availability in the coming weeks; that outlook does not set an Antigravity launch date. Google also announces OpenAI models for Gemini for business.

The move follows an [October 20 announcement bringing Claude Opus 5.5 and Sonnet 5.5 to Antigravity for Google Cloud customers](https://cloud.google.com/blog/products/ai-machine-learning/claude-opus-55-and-claude-sonnet-55-available-in-antigravity). In that case, Google explains how an administrator enables the models for developers to select. It is a concrete example of different model families meeting in the same tool, with their own enterprise access terms.

The confirmed change is more model choice within Google’s own tools. When Antigravity access arrives, the useful questions will be **which model is available, with what quota and at what price**. Those details will let you decide where it fits your work.

[Google Cloud’s official post on X](https://x.com/googlecloud/status/2112896659608785181).

<small>Editorial check: October 21, 2026. Sources: official announcements from [Google Cloud](https://cloud.google.com/blog/products/ai-machine-learning/google-cloud-brings-openai-models-to-customers) and [OpenAI](https://openai.com/index/openai-models-google-cloud/). The map summarizes those announcements; we did not verify access in an enterprise account or run performance tests. Official Google Cloud cover, kept complete and optimized for the web.</small>

## Plan your next post in HeyMark.

Keep the idea, review the draft with your team, and see how it performed in the accounts you connected.

[Start free](https://app.heymark.ai) · [See how it works](https://heymark.ai/en/#product)

## Developer and agent resources

- [HeyMark MCP documentation](https://heymark.ai/en/mcp/)
- [llms.txt](https://heymark.ai/llms.txt)
- [Full site content for language models](https://heymark.ai/llms-full.txt)
- MCP protocol endpoint: `POST https://mcp.heymark.ai`
- OAuth protected-resource metadata: [/.well-known/oauth-protected-resource](https://mcp.heymark.ai/.well-known/oauth-protected-resource)
- MCP server card: [/server-card](https://mcp.heymark.ai/server-card)