Google has introduced Gemini 3.1 Flash Live, a new voice model that improves real-time dialogue, task execution and multilingual interactions.
The company said the model offers faster response times, better tonal understanding and more reliable performance for developers, enterprises and general users.
The model is being rolled out across multiple products, including Gemini Live, Search Live and Gemini Enterprise. Developers can access it in preview through the Gemini Live API in Google AI Studio.
According to Google, Gemini 3.1 Flash Live scored 90.8% on ComplexFuncBench Audio, a benchmark that measures multi-step function calling.
On Scale AI’s Audio MultiChallenge, it recorded 36.1% with ‘thinking’ enabled, which evaluates long-horizon reasoning in real-world audio conditions.
“We’ve improved 3.1 Flash Live’s overall quality, making it more reliable for developers and enterprises to build voice-first agents that can complete complex tasks at scale,” the company said.
The model also improves tonal recognition, allowing it to detect pitch and pace variations, and respond to user emotions such as frustration or confusion. Google said this enables more adaptive conversations, particularly in enterprise customer service use cases.
Companies including Verizon, LiveKit and The Home Depot have tested the model and reported improvements in conversational flow within their workflows.
For consumers, Gemini 3.1 Flash Live powers updates to Gemini Live and Search Live. Google said the model delivers faster responses and can maintain conversational context for longer sessions.
It also supports multilingual interactions, enabling Search Live to expand to more than 200 countries and territories.
“With the 3.1 Flash Live model under the hood, Gemini Live delivers faster responses compared to the previous model, and it can follow the thread of your conversation for twice as long,” the company said.
Google added that all audio generated by the model is watermarked using SynthID, which embeds an imperceptible signal in the output to allow detection of AI-generated content.
ALSO READ: Infosys Acquires Optimum Healthcare IT and Stratus
