OpenAI's Astra, Sol and Luna are coming to Google Cloud, which now offers Gemini, Claude and GPT under one bill

The model you use is becoming a setting, not a vendor relationship, and your automations should be built that way.

What happened

OpenAI and Google announced that OpenAI's models are coming to Google Cloud. Ars Technica reports that Astra, Sol and Luna will appear in Google's Model Garden, which already lists more than 200 options including Gemini, Claude and open models like Mistral, Llama and Gemma. Google says it is now the only large cloud provider offering all three leading closed model families.

According to The Decoder, Astra targets demanding work like coding and research, Sol is a cheaper option with similar performance, and Luna is the fast, low cost choice for everyday use. The models launch in private preview for select enterprise customers, with sign ups open. Charges run through a single Google Cloud bill and count toward existing contracts, and Google says customer data will not be used for training.

Ars notes the models run entirely on Google's infrastructure, that Google's system supports moving workloads between models based on performance, latency and efficiency, and that GPT models will also appear in Google's Antigravity coding tool. OpenAI says Microsoft remains its primary cloud partner.

My take

For clients already on Google Workspace and Google Cloud, procurement was often the real blocker to using GPT models. A new vendor meant a new contract, new billing and a new security review. Putting all three model families behind one bill removes a lot of that friction.

The practical move is to design workflows so each step names its model in one place:

1. Classification, tagging and routing go to the cheap, fast tier.
2. Drafting customer facing text goes to a middle tier and gets a review step.
3. Only the hard reasoning steps get the expensive model.
4. Keep a test set of 20 to 50 real examples per step, so switching models is a measured decision.

Do not rush production work onto a private preview. Do start mapping which steps in your CRM and ops workflows are paying top tier prices for simple jobs. That audit usually pays for itself before any migration.

Not sure which workflow step is driving your AI bill? Message me through romielwillautomate.dev with the stack you run, and I will tell you where I would look first.