Google has announced Gemini 3.1 Flash Live, its latest audio and voice model designed for real-time conversations.
Google says the new model improves reliability when handling complex, multi-step tasks.
On the ComplexFuncBench Audio benchmark, it recorded a score of 90.8%, higher than its predecessor. It also scored 36.1% on Scale AI’s Audio MultiChallenge test, which measures instruction-following and reasoning in real-world audio conditions.
Gemini 3.1 Flash Live also introduces better tonal understanding.
It can detect variations in pitch, pacing, and user emotion, allowing it to respond more appropriately during conversations, including cases where users show confusion or frustration.
For general users, the model brings faster replies and longer conversation memory. Google says it can maintain context for up to twice as long as before, which may help during extended discussions or brainstorming sessions.
The model also supports multiple languages by default. With this, Search Live is expanding globally, enabling real-time voice interactions in over 200 countries and territories.
READ: Google Expands Search Live Globally
Google added that all audio generated by Gemini 3.1 Flash Live includes SynthID watermarking. This is meant to help identify AI-generated audio and reduce the risk of misinformation.
Gemini 3.1 Flash Live is now available starting today.
The model is now available across several Google platforms.
Developers can access it in preview through Google AI Studio, while businesses can use it via Gemini Enterprise for Customer Experience.
It is also rolling out to general users through Gemini Live and Search Live.
ManilaShaker is a tech media producing insightful and helpful content for our local and growing international audience. Our goal is to create a premier Philippine digital consumer electronics resource that provides the most objective reviews and comparisons globally.
