The short answer
● Released — google-adk 2.14.0, 7 November
● What changed — experimental features graduated to stable
● Now stable — execution environments, A2A protocol, evaluation, core runtime
● Reported gain — 23% faster long multi-turn workflows, 9% lower peak memory
● Languages — Python, TypeScript, Go, Java and Kotlin
Google released version 2.14.0 of its Agent Development Kit on 7 November. The release is a graduation rather than a feature drop: execution environments, the A2A protocol, the evaluation modules and the core runtime all move from experimental to stable, along with 26 auth, environment and app modules and 16 toolset features.
Stable A2A is the part that matters outside Google. Agent-to-agent communication has been a moving target in every framework that implements it, and a vendor marking its own implementation stable is a commitment to API shape that teams can build against.
New in the release: MultiRunConfig for concurrent multi-run evaluation, ToolArgsMatcher for semantic argument comparison, a TeamsRunner for Microsoft Teams, and an OpenAIDecisionsClassifier for OpenAI’s Decisions API. Reported performance is 23% faster on long multi-turn workflows with 9% lower peak memory, which is Google’s own figure and has not been independently verified.
Sources: PyPI · GitHub.
Published 8 November 2026. Every figure above was checked against the reporting linked here on the same day.