OpenAI’s GPT-6 models launch on Google Cloud, completing trio of major AI providers

Google Cloud announced on Wednesday that three of OpenAI’s newest frontier models—GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna—are now available in a private preview for select enterprise customers. The models join Google’s own Gemini line and Anthropic’s Claude within the platform’s Model Garden, making Google the only hyperscale cloud provider that hosts all three leading closed-source AI families.

• OpenAI’s GPT-6 suite is offered through Google Cloud Marketplace and will be billed on existing Google Cloud contracts.
• Customer data and generated content remain inside the client’s data perimeter and are not used to train the models.
• The integration also extends to Google’s Antigravity coding tool and Gemini Agent, with broader availability slated for the coming weeks.

How the partnership works

Google’s Model Garden already supports Anthropic’s Claude alongside Gemini. With the addition of OpenAI’s GPT-6 models, developers can select the most suitable model for a given workload from a single console. According to a Google Cloud blog post, purchases are processed through the Google Cloud Marketplace, allowing enterprises to apply existing spend commitments and receive a consolidated invoice.

Technical and business context

The deal follows OpenAI’s recent shift away from an exclusive license with Microsoft. After Amazon added the models to its Bedrock service in April, Google’s entry completes coverage of the three biggest U.S. cloud platforms. Google Cloud’s revenue growth accelerated to roughly 82 % in the second quarter, a surge that analysts attribute in part to strong demand for AI workloads and custom chips. By hosting OpenAI’s models on dedicated Google infrastructure, the partnership also addresses concerns about data residency and model training privacy.

Implications for developers and enterprises

For teams building AI-driven applications, the integration simplifies procurement and operational overhead. Developers can invoke GPT-6 models, Anthropic’s Claude, or Google’s Gemini via the same APIs, choosing between performance-oriented Astra, cost-effective Sol or fast-response Luna depending on the task. The models are also slated to appear in Google Antigravity, a low-code environment for generating code, and in the Gemini Agent, expanding the range of use cases from natural-language assistants to automated coding.

What comes next

OpenAI indicated that general availability will roll out in the “coming weeks,” though exact dates were not disclosed. Google’s CEO Thomas Kurian emphasized the goal of giving customers a broad selection of models to achieve optimal quality and lower costs. Analysts such as Mike Leone of Moor Insights note that Gemini now competes directly with OpenAI models on the same platform, raising the stakes for model performance and pricing.

Both companies remain competitors in search and AI research, but the partnership illustrates a pragmatic approach to reaching enterprise customers who already operate on Google’s cloud. As OpenAI continues to court investors and prepares for a potential IPO, broader distribution across all major clouds could become a key revenue driver.