Google announced Gemini 3.1 Flash Live on March 26, 2026, describing it as “our highest-quality audio and voice model yet,” according to both blog.google and deepmind.google.
According to the sources, the model features “improved precision and lower latency to make voice interactions more fluid, natural and precise.” The sources state the model is “faster and better at understanding tone for natural conversations” and is designed for natural and reliable real-time dialogue.
Developers can access Gemini 3.1 Flash Live through the Gemini Live API in Google AI Studio, while enterprises can implement it for customer experience applications, according to the announcement. The sources indicate that developers can “use it to build voice agents that handle complex tasks more reliably.”
According to both sources, the general public can experience the technology through Search Live and Gemini Live, which now support over 200 countries. The sources note that both services “now offer more helpful responses in many languages.”
A notable security feature is included: “All audio from 3.1 Flash Live is watermarked to help prevent the spread of misinformation,” according to the sources.
The announcement was authored by Valeria Wu, Product Manager, and Yifan Ding, Software Engineer representing the Gemini team, according to both blog.google and deepmind.google.