Gemini’s main developer path is now the Interactions API
Google has made the Interactions API generally available and is making it the main way to work with and agents. It first appeared as a public beta in December 2025, and the official release now gives it a stable structure. Developers can send a model ID to get model output, or send an agent ID to run a more independent task.
Long jobs can run in the background on Google’s servers. can create a remote Linux workspace where an agent can run code, browse the web, and manage files. Tool use is broader now, so , Google Maps, custom functions, and image results can be combined in one request.
A Flex tier is meant to cut cost by 50%, and paid users can retrieve past interactions for 55 days. The older generateContent API will keep working, but newer long-running and agent features are expected to move toward the Interactions API first, and sometimes only there.
Key points
- The Interactions API is now Google’s default API for and agents.
- can work inside a remote Linux workspace with code, web, and file access.
- lets long tasks keep running on the server.
- The Flex tier is presented as a way to reduce cost by 50%.
- The old generateContent API remains supported, but newer agent features are likely to center on the Interactions API.