What happened
Google announced the general availability of its Interactions API, now the primary interface for Gemini models and agents, following its public beta launch in December 2025. This API unifies access to Gemini models and agents, offering server-side state, background execution, and tool combination. New capabilities include Managed Agents for remote Linux sandbox execution, background processing for long-running tasks, and enhanced Deep Research with multimodal grounding. It also introduces Flex and Priority tiers for cost optimisation or latency, with 55-day retention on paid tiers.
Why it matters
Platform engineers and architects now have a unified, stateful API for agentic workflows, reducing integration complexity for Gemini deployments. The introduction of Managed Agents and background execution shifts operational burdens to Google, potentially lowering infrastructure costs for teams building autonomous applications. Cost optimisation tiers offer procurement teams flexibility, while the API's design for frontier capabilities suggests future model advancements will prioritise this interface, requiring migration for existing generateContent API users.
