OpenAI Models Are Coming to Google Cloud

Google Cloud customers get a new procurement path for OpenAI models, but the private preview still leaves production questions open.

The AI Enabled PM

Oct 22, 2026

OpenAI is bringing frontier models and Codex workflows to Google Cloud’s Model Garden. GPT-Astra, GPT-6.1 Sol, and GPT-6 Luna are in private preview for selected enterprise customers. Teams can use Google Cloud APIs, connect local Codex CLI and Desktop workflows, and buy usage through the Cloud Marketplace. Existing GCP commitments may cover the spend.

Tweet

For teams already on Google Cloud, this removes friction from procurement and early testing. A shared catalog does not make the models interchangeable. Use the preview to run one production-shaped agent workflow across OpenAI, Gemini, and Claude, then compare task quality, latency, unit cost, observability, tool behavior, and switching effort.

Several production questions remain open. Neither company has published pricing, regional coverage, quotas, or service levels. Google says prompts, context, and outputs stay inside the customer’s designated data perimeter and are not used to train foundation models. Zero Data Retention and Confidential Compute are still coming soon. Keep rollout gated on the controls and contract terms the product requires.

Sources

A working app with a success check balanced on unstable foundations.

Your Vibe-Coded App Works. Is It Ready for Customers?

A field guide for deciding when a vibe-coded app can move from internal testing to real customers.

An agent-shaped key approaches a product’s access-controlled doorway.

Your Next User Is an AI Agent

Products built for people need new identity, authority, observability and recovery controls when AI agents become active users.

A message leads to a bounded yes-or-no decision, shown as a physical switch.

Jev Shows That Not Every AI Call Should Generate Text

Jev returns bounded choices and probabilities instead of prose. Its real lesson for product teams is where generation should end and ordinary software should take control.

Overlapping conversation and reasoning bubbles: a live waveform and a brain.

Voice Agents Need Two Brains

Production voice agents need one loop for live conversation, another for reliable execution, and application-owned state between them.