OpenAI GPT 6.1 Models Head to Google Cloud and Antigravity

Summary

OpenAI’s GPT 6.1 suite, Astra, Sol and Luna, is entering Google Cloud, despite Microsoft remaining OpenAI’s primary cloud partner. On October 21, 2026, Google said the models were rolling out in private preview to select enterprise customers before reaching Model Garden in the coming weeks. The catalog already exceeds 200 options, including Gemini, Mistral, Llama, Gemma and Anthropic’s Claude, and Google says it is now the only major cloud provider carrying all three leading closed LLM families.

Customers can run GPT inference entirely on Google infrastructure and buy it through Google Cloud Marketplace, letting companies apply existing Google Cloud spending commitments. Early-access requests unlock API use, including through Codex CLI and Codex Desktop. Google can shift workloads dynamically among LLMs for performance, latency and efficiency, although Astra will be the costliest GPT 6.1 model to run. GPT is also coming to Google’s agent-powered coding client Antigravity, which currently supports Gemini and Claude, but Google has not confirmed whether Astra, Sol and Luna will all appear. The expansion adds another option after Gemini 4 Argon, which Google says restores competitive parity and improves on Gemini 3.1 Pro, but consumes tokens very quickly.