Google Cloud announced “Gemini Agent,” a cloud-based multi-agent tool, on October 9, 2026 (local time). In addition to Google’s new “Gemini 4 Argon” model, users can choose Anthropic’s “Claude Opus 5.5” and “Claude Sonnet 5.5.” Other closed and open models are expected to be added over time.
Claude runs under the Gemini name
One striking aspect of the announcement is that Claude is available through a service bearing the Gemini name. Gemini Agent is designed not to be tied to any particular model, and says it will use the model best suited to each task.
To clarify, AI models are the brains responsible for generating text and reasoning. Agents are systems that use those brains to carry out actual tasks. Gemini Agent is a tool that brings together multiple agent systems. Its role is easier to understand if you distinguish between the foundation that runs the tasks and the models used within it.
Until now, choosing a service has often meant choosing its model as well. With Gemini Agent, users can choose a model after selecting the service.
Three models available now, with more to come
The three models currently available are Gemini 4 Argon, Claude Opus 5.5, and Claude Sonnet 5.5. Google’s Shubham Sabu revealed this on X alongside a screen showing the model selection options, and said other closed and open models are expected to be added in the future.
The prospect of Google’s cloud offering models from outside Google shifts the focus of the discussion away from which model is superior and toward how to use them—which model to assign to which task.
How will this change the way businesses work?
If multiple models can be accessed from one place, it becomes easier to match them to the nature of each task—for example, using a faster model to draft documents and a more advanced model for complex reasoning. The reduced need to build workflows around a single model is a significant change for companies considering adoption.
However, being able to choose a model is not the same as being free to switch services. The announcement did not specify the terms for connecting to or migrating from other services. Companies considering adoption may find it easier to assess the offering if they think concretely about how they would use different models across their own operations, in addition to which models are supported.
How it relates to existing Gemini apps and pricing
Google has not yet clarified how Gemini Agent relates to the existing Gemini app or Gemini Enterprise. Nor did it announce that Claude in the cloud can now be used directly in the Gemini app.
It is also unclear which plans will include the service: Gemini AI plans for individuals, Google Workspace or Gemini Enterprise for businesses, or some other option. Google Cloud Japan has been asked about pricing and plans for availability in Japan; a follow-up is expected once it responds.
Comments