OpenAI's GPT-6 models are coming to Google Cloud

Two of the fiercest rivals in AI are now doing business together. On 21 October OpenAI and Google Cloud announced that OpenAI's newest models are coming to Google Cloud's model garden, its catalogue of AI models that businesses can call through Google's APIs.

What's actually launching

• Models: GPT-6 Astra (the top model, for coding, computer use and research), GPT-6.1 Sol ("near-Astra" ability at lower cost) and GPT-6 Luna (fast and cheap).
• Who gets it: a private preview for select enterprise customers, starting today. OpenAI says general availability is planned "in the coming weeks."
• Billing: through Google Cloud Marketplace, on the same Google invoice, and it counts against existing Google Cloud spending commitments.
• Codex: developers can point the local Codex CLI and desktop app at Google's platform for inference.
• Later: OpenAI models are also coming to Google's Antigravity coding tool and to Gemini for business.

Google promises that prompts and outputs stay inside the customer's data perimeter and aren't used to train models, and that OpenAI models run on dedicated Google Cloud infrastructure. OpenAI says Microsoft remains its "primary cloud partner." As CNBC points out, Amazon began hosting OpenAI models in the spring after Microsoft's exclusive licence ended, so this deal puts OpenAI on all three of the biggest US clouds.

Our take

For most people this changes nothing in ChatGPT. It matters for companies, because the hardest part of adopting AI at work is often procurement, not picking a model: a new vendor contract, a security review and a separate bill. Selling GPT-6 inside Google's marketplace and against existing cloud budgets removes most of that.

Google gains as well. It now calls itself the only hyperscale cloud offering all three leading closed-source model families (Anthropic's Claude was already there next to Gemini). That makes Google Cloud look like a neutral shop rather than just Gemini's home. The cost is that Gemini now sits beside its strongest competitor in Google's own catalogue.

The real competition is moving from benchmark charts to distribution. That echoes this week's chip news: Google and Amazon have both opened up PyTorch tooling to win developers. If you'd rather keep AI on your own machine, our guide to choosing a laptop for local AI is a good place to start.