OpenAI Brings GPT-6 Astra, Sol and Luna to Google Cloud in a Private Preview

OpenAI and Google Cloud announced on October 21, 2026 that OpenAI's frontier models and Codex workflows are coming to Google Cloud's Model Garden. Google says GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna are rolling out now in a private preview for select enterprise customers, and OpenAI says general availability is planned in the coming weeks.

The deal puts OpenAI's models on all three of the largest US cloud platforms, after Amazon started hosting them in April (CNBC). It is the latest of OpenAI's enterprise products and tools to reach customers through a rival's platform: Google competes with OpenAI through its own Gemini models, and Microsoft, a major OpenAI shareholder, stays its primary cloud partner, OpenAI said in its announcement.

Which OpenAI models are coming to Google Cloud?

Google Cloud says three OpenAI models are rolling out in its Model Garden from October 21, 2026: GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna. OpenAI's own announcement refers to its "frontier models" and Codex without naming versions.

Google's blog post, written by Michael Gerstenhaber, vice president of product management for Gemini at Google Cloud, positions each model for a different job:

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

Ars Technica calls this the full suite of current GPT-6 and GPT-6.1 models. Google says customers can swap between models based on performance, latency and cost without separate vendor contracts. Developers can call the models through Google's APIs and, according to Google, connect their local Codex CLI and Codex Desktop setups so that inference runs on Google Cloud. Codex is OpenAI's coding agent.

How do billing and data handling work?

Both companies say customers buy OpenAI models through Google Cloud Marketplace, so enterprises can pay for them from the spend they already committed to Google Cloud. Google adds that all usage appears on a single Google Cloud invoice and counts toward its Flexible Savings Plans across model providers.

On data, Google says customer prompts, context and outputs stay inside the customer's data perimeter and are never used to train foundation models. The company says the OpenAI models run on dedicated Google Cloud infrastructure behind isolated gateways, with VPC Service Controls and unified audit logging. Two protections are only promised: zero data retention options and Confidential Compute are both listed as "coming soon". These are Google's descriptions of its own setup; no independent review was published with the announcement.

Box is the only customer quoted in Google's post. Yashodha Bhavnani, Box's vice president of AI products, said in Google's post that the company will be able to offer OpenAI models to its customers with Google Cloud's security and compliance controls.

Why can OpenAI sell through Google now?

OpenAI can sell through Google because Microsoft and OpenAI amended their partnership in April 2026, ending Microsoft's exclusive rights to sell OpenAI's models. Under the new terms, reported on April 28 (ANI via The Tribune), Microsoft keeps a non exclusive license to OpenAI's models and products through 2032, and OpenAI products still launch first on Microsoft Azure unless Microsoft chooses not to support a capability.

Amazon moved first. It unveiled plans to host OpenAI models the day after the exclusivity ended (CNBC), and Google now follows. The two rivals have worked together before: OpenAI said in 2025 that it was using Google's cloud to run ChatGPT and its API, according to CNBC.

Read together, the April terms and the October announcement suggest what Google is getting: the current GPT-6 lineup on its platform, while new OpenAI products are still due on Azure first. Neither company addressed launch timing in its October 21 post.

What changes for Antigravity and the Gemini agent?

OpenAI says its models "will soon be available in Google Antigravity and the Gemini Agent," without a date. Google's post names Antigravity and Gemini for business instead; neither company has said which of the three GPT-6 models those products will get, Ars Technica noted.

Antigravity is Google's agent based coding tool, and until now it has supported Google's Gemini models plus Anthropic's Claude, according to Ars Technica. Google announced Claude Opus 5.5 and Claude Sonnet 5.5 for Antigravity in a separate post. The Gemini agent, which Google Cloud unveiled on October 8, 2026, already routes work across Gemini and Claude models. Once OpenAI's models arrive, both of Google's own agent tools would offer models from all three makers of what Google calls the leading closed source LLMs, alongside Gemini 4 Argon.

Greg Brockman, OpenAI's president and co-founder, said in the same announcement that the partnership brings OpenAI's most capable models and Codex to where customers already build.

What we don't know yet

• Per token prices for GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna on Google Cloud, and how they compare with OpenAI's own API.
• Which Google Cloud regions will host the models, including availability for customers in the EU.
• The date of general availability and the dates for Antigravity and the Gemini agent.
• Whether future OpenAI models will reach Google Cloud on the day they launch or only after Azure.