OpenAI's GPT-6 models arrive in Google Cloud's model garden

Google Cloud announced on 21 October 2026 that OpenAI's GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna are rolling out in private preview to select enterprise customers in its model garden. Michael Gerstenhaber, Google Cloud's VP of product management for Gemini, wrote that Google Cloud is now the only hyperscale cloud provider offering all three leading closed-source language models, alongside more than 200 first-party, third-party and open models. That is Google's claim about its competitors, not something I could check.

What is on offer

• GPT-6 Astra: described by Google as designed for demanding work across coding, computer use, professional tasks and scientific research.
• GPT-6.1 Sol: "near-Astra intelligence at lower cost" across agentic coding, computer use and professional work, in Google's words.
• GPT-6 Luna: a fast, cost-efficient model for everyday work at scale.

Google says customers can call the models through its APIs and connect their local Codex CLI and desktop workflows to the Gemini platform for inference. Access is by form: Google's sign-up page says filling it in "does not guarantee access". Google did not publish per-token prices, rate limits or Regions in the post.

What changes for a buyer

The commercial change is the larger one. Google says OpenAI models are transacted directly through Google Cloud Marketplace, so usage appears on one Google Cloud invoice, draws down existing cloud commitments and counts toward Flexible Savings Plans across model providers. For a team that already holds a Google Cloud commitment, that could mean trying OpenAI models without a new vendor contract, though you would need Google's price list to know what that costs.

Google also describes its controls: prompts, context and outputs stay inside the customer's data perimeter and are never used to train foundation models; requests go through isolated gateways; and every agent and model interaction runs under an authenticated identity with VPC Service Controls and unified audit logging. Box's VP of AI Products, Yashodha Bhavnani, is quoted in the post saying Box will use the models to give customers access to OpenAI's models alongside Google Cloud's security and compliance.

What is not live yet

Google's post lists Zero Data Retention options and Confidential Compute for OpenAI models as "coming soon", and says the models are coming to Gemini for business and Google Antigravity "soon". The Decoder's short report says the models will also appear in those two products, without a date. Treat Google's post as the record.

What to do

If you run on Google Cloud, submit the interest form with a named workload and a test set, not a general request. Ask your account team three things in writing: the per-token price against OpenAI's own API, whether the data-retention terms you need are live or still promised, and which Regions serve the models. Until those answers arrive, keep any production routing on the endpoints you already trust.