Google’s Interactions API reaches general availability, becomes primary interface for Gemini models and agents

Google said Monday that its Interactions API has reached general availability and is now the company’s primary interface for working with Gemini models and agents, marking a clear product shift toward a more agent-focused development path. In a blog post, Ali Çevik, a group product manager at Google DeepMind, and Philipp Schmid, a developer relations engineer at Google DeepMind, wrote: “Today we're announcing that the Interactions API has reached general availability and is now our primary API for interacting with Gemini models and agents.”

For developers, the change matters because Google is making Interactions the default interface across Google AI Studio, the Gemini API and its documentation. The older generateContent endpoint is not going away, and Google said it will continue to be supported and will keep receiving mainline Gemini models. But the company is directing new projects toward Interactions and said frontier capabilities for long-running models and agents will increasingly launch only on that API.

Google first introduced the Interactions API in public beta in December 2025. With Monday’s general release, the company said the API now has a stable schema, a sign that the structure developers code against is no longer expected to shift in major ways. Google also published a migration guide for developers moving from generateContent, underscoring that the announcement is as much about transition planning as it is about new features.

The Interactions API is designed as a more unified way to handle multi-step AI workflows, rather than a simpler one-shot text generation call. Google highlighted several capabilities at GA, including Managed Agents, background execution for long-running tasks, tool upgrades, Deep Research improvements, media generation and a schema simplification it described as moving “From Roles to Steps.” Developers can now set background=True to run tasks asynchronously on Google’s servers. Google also said the API now better supports mixing built-in tools with user-defined functions and allows tool results to return images.

A central piece of the launch is Managed Agents, which package reasoning, execution and tool use into a server-side environment. Google said: “A single API call provisions a remote Linux sandbox where an agent can reason, execute code, browse the web and manage files.” The default managed agent is called Antigravity, though developers can also define custom agents. Google also pointed to Deep Research upgrades, including two agent versions tuned for speed versus depth, collaborative planning, native charts and infographics, and multimodal grounding with images, PDFs and audio. On media, Google highlighted image generation with Nano Banana 2, music generation with Lyria 3, and multi-speaker expressive text-to-speech.

Google said the Interactions API includes Flex and Priority service tiers, with Flex offering a 50% cost reduction. The company also said past interactions are retrievable, with 55-day retention on the paid tier. The API is available through Google’s official Python and JavaScript software development kits, as well as through Google AI Studio, where developers can access API keys and documentation.