OpenAI Lands on Google Cloud: Picking a Cloud Just Got Easier
In short
- On October 21, OpenAI announced that its frontier models and Codex agentic workflows are coming to Google Cloud’s Model Garden.
- It starts as an invite-only preview, with general availability planned “in the coming weeks”. You pay through Google Cloud Marketplace, and the spend counts against your existing Google Cloud commitments.
- Money angle: if your company already runs on Google Cloud, you can now use GPT models and Codex without a second cloud contract or a second invoice.
OpenAI is now on all three big clouds. First Microsoft Azure, then Amazon in the spring, and now Google Cloud. That means GPT models and Codex inside the platform of OpenAI’s biggest competitor. If you’re picking a cloud, the rules just changed.
What happened
On October 21, OpenAI announced that its frontier models are coming to Google Cloud’s Model Garden – the catalog where Google already lists Gemini, Claude and open models. Codex, OpenAI’s coding agent, comes too, with its agentic workflows.
Can you use it today? Not yet, unless you’re invited. It starts as a limited preview for select customers. General availability is planned “in the coming weeks”, with no exact date.
How do you pay? Through Google Cloud Marketplace. The usage goes on your normal Google invoice and draws down the commitments you’ve already signed. No new vendor, no new procurement process.
And Microsoft? OpenAI says Microsoft stays its primary cloud partner. But the exclusive era ended in April, when the two companies rewrote their deal. A day later, OpenAI models landed on Amazon Bedrock. Google is simply the third.
How to use it
- Already on Google Cloud? Ask your account manager about the preview. You can test GPT next to Gemini on the same data, with the same access controls.
- Pick the model per task. One platform, several vendors. Use the cheaper model for routine work and the strongest one only where quality pays for itself.
- Dev teams: Codex arriving on Google Cloud means you no longer have to move to another cloud just to get OpenAI’s coding agent.
Do the math: imagine a company that pays OpenAI’s API €1,500 a month on a separate card, plus a Google Cloud commitment it already pays for. That’s €18,000 a year on a second invoice. Once the same usage runs through Marketplace and counts toward the Google commitment, the separate bill can disappear. That doesn’t make the tokens cheaper. It’s about using a budget you’ve already committed. (Illustrative example – Google has not published per-token prices for OpenAI models yet.)
What it means
The model is no longer the reason to choose a cloud. Before, if you wanted GPT, you leaned toward Azure. Now you pick a cloud on price, data location and the tools your team knows. The models follow you.
Microsoft loses its strongest selling point. It still has the deepest partnership, but “only here” is gone. Google wins: Gemini and GPT on one shelf means customers have one less reason to leave. And OpenAI wins most. It sells everywhere and gets more leverage with every partner.
The weak spot? No public price list and no launch date yet. Until those exist, this is a promise, not a product.
Ivan’s Take
For businesses, this is good news, plain and simple. Lock-in is expensive, and now the best models live on every big cloud. My advice: don’t migrate anything because of a model. Stay where your data and team already are, and once GA lands, test GPT and Gemini side by side on your own tasks before you commit.
Source: OpenAI
If it doesn’t save you time or money, you won’t see it here
- Facts, not hypeEvery price and number is checked against the official source
- Always the real priceYou see what it costs before you click
- I say when it’s weakEven when it’s OpenAI or Google
