Google's Interactions API for Gemini Reaches General Availability

Today, Google announced that the Interactions API has reached general availability and is the primary API for interacting with Gemini models and agents. [1] The interface launched in public beta in December 2025 and now features a stable schema. [1] Google has defaulted its documentation to the Interactions API and is working to make it the default across third-party SDKs and libraries. [1]

The release adds new capabilities including background execution, which allows the server to run interactions asynchronously by setting "background=True" on any call. [1] Google also introduced Managed Agents, which use a single API call to provision a remote Linux sandbox where agents can manage files, execute code, and browse the web. [1] The default managed agent is called Antigravity, and developers can define custom agents using their own instructions and data sources. [1] Additionally, the update includes tool improvements that allow developers to mix built-in tools like Google Search and Maps with their own functions. [1]

The elevation of the Interactions API to a primary, unified endpoint simplifies the developer experience for Gemini. By standardizing background processing and managed agents—including a direct remote Linux sandbox—Google is lowering the friction required to deploy complex, autonomous workflows. This compares directly to OpenAI's Assistants API, centralizing inference, tooling, and environment provisioning into a single programmatic interface. What the sources don't address: How the pricing structure will change for long-running background tasks and managed agent sandbox compute usage.