OpenAI and Google Cloud have partnered to offer 3 GPT-6-series models on Model Garden. Codex integration and enterprise deployment are progressing. The 3 models currently in private preview will be available via API, while cloud procurement integration and data-protection features are also being expanded.
OpenAI and Google Cloud announced a partnership that will make OpenAI's advanced AI models and Codex's agentic workflows available on Google's Model Garden. This is expected to allow enterprises to use OpenAI's technology in their existing Google Cloud environments for software development and complex multistep tasks.
The initial rollout includes 3 models, GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna, which are currently available to selected enterprise customers in private preview. Approved customers can connect via API and run inference by connecting the local Codex CLI or desktop workflows directly to Google's Gemini platform. General availability is expected within weeks.
Enterprises will be able to select and switch to the appropriate model based on requirements such as performance, response speed and cost. Procurement and billing can be consolidated through Google Cloud Marketplace, and cross-provider usage will be available through Flexible Savings Plans (FSP).
Data protection is also a focus. Customer prompts, context and outputs will be retained within the data boundary and will not be used to train foundation models. OpenAI's models will run on dedicated infrastructure, combined with controls such as authentication and audit logs. Options including Zero Data Retention (ZDR) and additional confidentiality features are also planned. The partnership is expected to expand to Gemini for business and Google Antigravity, and the announcement also specified that Microsoft remains a major cloud partner.