Google Cloud and OpenAI have entered into a partnership to bring OpenAI’s frontier models and Codex into Google Cloud’s Model Garden. This integration allows enterprise customers to access OpenAI’s models within the Google Cloud environment, combining OpenAI’s capabilities with Google’s high-performance infrastructure and data governance tools. The move is designed to help organizations build and scale complex agentic workflows—automated systems that can perform multi-step tasks—while maintaining the security standards required by regulated industries.
This image illustrates the primary partnership between OpenAI and Google Cloud to bring frontier models into the Model Garden.
The integration provides specific access to a range of OpenAI models, including GPT-6 Astra for demanding professional tasks, GPT-6.1 Sol for cost-efficient agentic coding, and GPT-6 Luna for high-volume everyday work. Customers can purchase these models through the Google Cloud Marketplace, which allows them to draw down existing Google Cloud Platform (GCP) spend commitments and receive unified invoicing. Furthermore, the models will eventually be integrated into Google Antigravity and the Gemini Agent, offering users the ability to swap between different models based on specific needs for latency, cost, or performance without requiring separate vendor contracts.
To address enterprise security concerns, the partnership includes strict data perimeters where customer prompts and outputs remain within designated boundaries and are not used to train foundation models. Google Cloud plans to offer Zero Data Retention (ZDR) options and will run OpenAI models on dedicated infrastructure with Confidential Compute features coming soon. While Microsoft remains OpenAI’s primary cloud partner, this expansion into Google Cloud reflects a broader industry trend toward interoperability and providing customers with a diverse selection of models to optimize for different use cases. General availability for these models is expected to begin in the coming weeks.
This visual supports the section regarding data perimeters and security measures for enterprise customers.