OpenAI and Google Cloud announced a partnership to host GPT-6 models and the Codex agent on Google's platform for enterprise customers in private preview.
Key points
- GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna are available in private preview on Google Cloud Model Garden.
- OpenAI plans general availability for these models and the Codex agent in the coming weeks.
- Customers can use existing Google Cloud spending commitments to pay for usage, with billing consolidated on one invoice.
OpenAI and Google Cloud announced a partnership on Wednesday to host OpenAI’s frontier models and coding agent on Google’s infrastructure. This move allows enterprise customers to access GPT-6 variants and Codex through Google Cloud’s Model Garden in private preview.
What happened
OpenAI is rolling out GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna in private preview on Google Cloud, according to statements from both companies on Wednesday. Select enterprise customers receive access first, with general availability planned for the coming weeks. This partnership places OpenAI’s tools on all three major US cloud providers, following Amazon Web Services’ hosting in April after Microsoft’s exclusive license ended.
Google said customer data stays within the customer’s data perimeter and is never used to train models. The OpenAI models run on dedicated Google Cloud infrastructure. Customers purchase access via the Google Cloud Marketplace and can apply existing spending commitments. All usage appears on a single Google Cloud invoice, simplifying billing for enterprises.
Microsoft remains OpenAI’s primary cloud partner, OpenAI stated. Google now offers all three leading closed models, including Anthropic’s Claude and its own Gemini, alongside OpenAI’s offerings. Thomas Kurian, chief executive of Google Cloud, said the goal is to give customers a broad selection to choose the right model for any job.
Why it matters
Hosting multiple AI models on one platform allows organizations to compare performance and costs without switching providers. It reduces vendor lock-in by letting teams test different models for specific tasks. This competition among providers often leads to better pricing and improved service levels for customers.
The fix
- Monitor official announcements from Google Cloud and OpenAI for general availability dates.
- Review current cloud spending commitments to understand how new model usage will be billed.
- Test private preview access with non-production workloads to evaluate performance.
- Verify data handling policies to ensure compliance with internal security standards.
Why it matters
Organizations should evaluate whether consolidating AI services on one cloud platform simplifies their operations. Teams usually check if the new models integrate with existing tools before committing resources. It is important to verify that data privacy guarantees align with corporate security policies. Users often compare performance across different providers to ensure they are getting the best value for their specific use cases.
What to do and how to stay safe: OpenAI
- Check if your organization is part of the private preview program for early access.
- Review billing settings in Google Cloud Marketplace to track usage costs accurately.
- Ensure your development teams understand how to connect Codex to Google’s platform.
- Monitor for updates on general availability to plan for broader deployment.
Step-by-step guide: What is password policies and how they protect systems
General security guidance from the Crypt Weekly newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
Which OpenAI models are available on Google Cloud?
GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna are available in private preview.
When will general availability launch?
OpenAI plans general availability in the coming weeks.
How does billing work for these models?
Customers pay through Google Cloud Marketplace using existing spending commitments, with all usage on one invoice.
