Google has introduced a new model designed for real-time voice conversations, Gemini 3.1 Flash Live. It is now being rolled out to Gemini Live and Search Live. The update is accessible to all users and aims to improve response delays and conversation tracking issues from the previous version.

What Has Changed in Gemini 3.1 Flash Live?

Google states that the new model provides faster responses during live voice sessions and can maintain conversation context for twice as long as the previous version. The extended context window is designed to keep longer sessions consistent and allows for continuation of conversation without losing previous parts.

The model also adds acoustic awareness, meaning it can detect changes in a user's tone and speed during speech. According to Google, this allows Gemini Live to understand when a user is angry or confused and adjust its tone accordingly.

Google notes that the update specifically addresses situations where the previous model struggled; for instance, moments when users hesitate, stutter, or try to interrupt an ongoing response. Technical details on how the interruption process has been restructured have not been provided.

Search Live Expansion and Accessibility of Gemini 3.1 Flash Live

Along with the model update, Google is expanding Search Live to over 200 countries and regions this week. The expansion is made possible by Google's multilingual design, which allows for real-time multimodal conversations in the user's preferred language without the need for a separate localized version.

Google did not clarify which languages are supported at launch or whether all markets will have full feature parity from day one.

Gemini 3.1 Flash Live is currently being rolled out. Google has not provided a timeline for when the rollout will be completed across all regions and platforms.