Google Unveils Gemini 3.1 Flash Live: Largest Ever Upgrade for Gemini Live

Google has officially launched the Gemini 3.1 Flash Live voice model, calling it its "highest quality audio and voice model to date," which brings major upgrades to Gemini Live and Search Live services. The model is currently available for preview in Google AI Studio through the Gemini Live API, signifying important progress in Google's real-time voice interaction technology.

Compared to the previous generation 2.5 Flash Native Audio, Gemini 3.1 Flash Live more effectively recognizes acoustic details like pitch and rhythm, while reducing latency. It significantly improves environmental noise filtering, better distinguishing speech from background sounds such as traffic or TV. In the Gemini Live app on Android and iOS, users will get faster responses with "reduced awkward pauses," and conversation thread duration is doubled, ensuring coherent thinking during extended brainstorming. Additionally, Gemini Live can now dynamically adjust response length and tone to match the context.

Google is also using this model to launch Search Live service in over 200 countries worldwide, covering all languages and regions where AI Mode is available, promoting the global deployment of real-time voice search. This upgrade not only optimizes user experience but also demonstrates the deep integration potential of voice AI in practical scenarios.

The release of Gemini 3.1 Flash Live reflects Google's ongoing breakthroughs in voice model performance and real-time interaction, potentially further driving smart assistants toward more natural and efficient evolution, and accelerating the widespread adoption of voice technology globally.