Google Cloud is adding OpenAI models to Model Garden in a private preview for selected enterprise customers. Local Codex workflows can use Google’s platform for inference, and purchases pass through the existing cloud marketplace. Google describes common identity and audit controls; zero data retention options and support inside Antigravity remain planned additions.
Google Cloud has begun offering OpenAI models to selected enterprise customers in a private preview. The models join Model Garden, Google's catalogue for accessing its own and other providers' models. Organizations can request access, while the initial rollout remains limited. The arrangement puts OpenAI alongside Gemini and Claude on the platform.[1], [2]
The preview includes GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna. Applications can call them through an application programming interface, or API. Codex CLI, the command-line development tool, and Codex Desktop workflows running locally can also connect to Google's Gemini platform for inference, the calculation that produces a model's response. Transactions use Google Cloud Marketplace, with usage on the customer's cloud invoice and eligible existing spending commitments.[1]
Google says prompts, context and outputs stay within the designated data perimeter and are not used to train foundation models. Its stated controls include authenticated identities, isolated gateways and unified audit logging. OpenAI models run on dedicated Google Cloud infrastructure, according to the company. Zero Data Retention options, Confidential Compute and support inside Gemini for business and Antigravity are planned for later stages.[1]
