Google Introduces Gemini 3.1 Flash Live Audio Model
Google has announced the launch of Gemini 3.1 Flash Live, a new audio and voice AI model built to support real-time conversations with greater accuracy and reduced latency. The model is currently available in preview for developers through the Gemini Live API in Google AI Studio, for enterprises via Gemini Enterprise for Customer Experience, and for consumers through Search Live and Gemini Live.
Strong Benchmark Performance Highlights Advanced Capabilities
Gemini 3.1 Flash Live has demonstrated strong performance across key industry benchmarks. It achieved a score of 90.8% on ComplexFuncBench Audio, which evaluates multi-step function execution under constraints. Additionally, on Scale AI’s Audio MultiChallenge—focused on complex instruction handling and long-duration reasoning in real-world audio environments—the model reached 36.1% when its advanced “thinking” mode was enabled.
Enterprise Adoption and Real-World Feedback
Several major companies, including Verizon, LiveKit, and The Home Depot, have already reported positive results after integrating the model into their workflows. Gemini 3.1 Flash Live enhances tonal recognition, allowing it to better detect nuances such as pitch and speech pace. It can also adapt responses dynamically based on user emotions, including frustration or confusion, improving overall interaction quality.
Improved User Experience and Global Expansion
In consumer-facing applications, Gemini Live now delivers faster responses compared to previous versions and can maintain conversational context for significantly longer periods. The new model also supports the continued global rollout of Search Live, which is now accessible in over 200 countries and territories with multilingual functionality.
Built-In AI Watermarking for Content Transparency
To address concerns around AI-generated content, all audio produced by Gemini 3.1 Flash Live includes SynthID watermarking. This invisible marker is embedded within the audio output, enabling reliable detection of AI-generated material. Google emphasized that this feature is designed to help combat misinformation and improve transparency.
