Google Cloud Offers AI Models from Three Major Companies on a Single Platform

Google Cloud has become the first major cloud service provider to offer AI models from three leading companies—its own Gemini, Anthropic's Claude, and OpenAI's GPT models—on a single platform. This enables enterprises to choose among the three companies' models based on their specific needs without having to navigate multiple cloud environments.

For a long time in the cloud market, each provider has deployed services built around its own proprietary AI models. AWS has made substantial investments in Anthropic, while Microsoft leverages its close relationship with OpenAI to provide GPT models through its own cloud service, Azure. In this context, Google has adopted a strategy of offering Gemini while actively incorporating models from other providers, and today's move can be seen as a continuation of that approach.

At launch, the OpenAI models being offered are three variants: GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna. All are currently available as private previews limited to enterprise customers. A private preview refers to a phase in which specific companies and developers gain early access before general availability, representing a pilot stage leading to official release.

The significance of this move extends beyond simply adding another model. When enterprises integrate AI into their operations, a broader range of model choices enables more flexible design tailored to task characteristics, cost, and security requirements. Having a single access point through Google Cloud to multiple cutting-edge models could simplify operational management, particularly for enterprises undertaking large-scale system development.

Notably, how competing cloud providers respond will also be worth watching. The combination of offering OpenAI models through Google's platform would have been unthinkable under previous industry conventions. The shift reflects a growing demand from enterprise users to access the best models regardless of location—a need that now appears to outweigh cloud providers' traditional model lock-in strategies.

For now, availability remains limited to enterprise customers in private preview. The next focal points will be whether this transitions to general availability and whether the range of supported models expands. Should cloud platforms come to function as a shared infrastructure for AI models, competition over the deployment platform—where models are used—could become as important as competition over model development itself.