Gemini 3.1 Flash Live updates Google’s real-time voice stack

Google introduced Gemini 3.1 Flash Live on 26 March for real-time dialogue, with preview access through the Gemini Live API in AI Studio. The model also powered consumer experiences in Search Live and Gemini Live, while enterprises could use it through Google’s customer-experience offering.

Google reported improvements in function calling, handling interruptions and understanding changes in tone. It said Gemini Live could follow a conversation for twice as long as before, while Search Live expanded to more than 200 countries and territories. Those statements describe the rollout and vendor evaluations; they do not prove equal performance across accents or noisy environments. Generated audio includes SynthID, giving the output an origin marker without turning the conversation into an authoritative source.